CA1128645A - Transmission method and system for facsimile signal - Google Patents

Abstract

S P E C I F I C A T I O N

TRANSMISSION METHOD AND SYSTEM FOR FACSIMILE SIGNAL

ABSTRACT OF THE DISCLOSURE
A transmission method for a facsimile signal by the use of the two-dimensional coding principle, in which when succes-sively coding addresses of a facsimile signal representative of the positions of information change picture elements, each having a binary value different from that of an immediately preceding picture element, the above-mentioned addresses on each coding scanning line are classified into three modes that are determined by the states of information change picture elements on the coding scanning line and on a reference scan-ning line immediately preceding the coding scanning line.
The above two-dimensional coding principle and a one-dimensional coding principle may be adaptively adopted to shorten the trans-mission time and to lessen the influence of a transmission error.

Description

~2~

This invention relates to a transmission method for efficient transmission of a binary signal, such as a two level facsimile signal.
Heretofore, there have been proposed, as a two-level facsi-mile signal coding system, (l) a run-length coding system in which a signal obtained by scanning is converted in~o a time series train and then the magnitudes o~ the run lengths of white and black are successively coded alternatel~y with each other for transmission and ~2) a system in which signals of plural, for example, two scanning lines are simultaneously coded all together.
The system (1) does not utilize at all the property that facsi-mile signals have a high correlation in a direction perpendicular (vertical) to the scanning line direction ; therefore, the compression efficiency is low. The system (2) makes use of the correlation in the vertical direction with respect to the signals of a set of scannins lines to be coded at a time~but does not utilize the correlation to signals of this syst~m other scanning lines ; consequently, the compression efficiency is higher than that of the system (l) but no sufficient compression effect is achieved, since this system does not fully use the two dimen-sional correlation among adjacent scanning lin~s.
An object of this invention is to overcome such defects of the prior art systems and to provide a transmissio~ method u~ing a two-dimensional successive coding system which removes redun-dancy of a facsimile signal by a relatively small nu~her ofmemories and a simple circuit or means to thereby permit a sub-stantial reduction of the number o~ bits to be se~t outO
Another object of ~his invention,is to pr~ide a transmis-sion method using a one-di~ensional, two dimensional adaptive - ~L2~S

coding method in which the two dimensional successive coding principle and the one-dimensional coding principle, such as a run-l~ngth coding system, are adaptively adopted~ so that the amount of information or signals to be transmitted is reduced, thereby to shorten the transmission time and to lessen the influence of a transmission error.
Further object o~ ~his invention is to provide a decoding system suitable for decoding a facsimile signal coded by the above mentioned coding method.
The present invention relating ~o the first object is based on the principle that when sucressively coding information (hereinafter referred to as addresses) o a facsimile signal representative of the positions of informa~ion change picture elements (hereinafter referred tD simply as change picture ele-ments), each having a binary signal ~alue dif~eren~ from that of an immediately preceding piGture element, the~number of pic-ture elements ~hereinafter referred to as the distan~e) between each change picture element to be coded and a selected one of the adjoining change picture elem~nts on the same scanning line (hereinafter referred to as a co~ing line) as the change picture element to be coded or on a scanning line im~ediately preceding it (which scanning line will hereinafter be ref~rred to as a reference line) is employed to be classiied into ~hree mQde~
determined by the combinations o~ states of the above inf~r-mation change picture elem~nts.
The present invention relati~g to the sec~d objec~ isbased on the principle that in the coding of a digital facsimile signal, picture signal information o~each line is coded by the one-dimensional system (for example, a run-leng~h coding system) .

~2~

and the two-dimensional system and, for each line, the t~o coded signals are compared with each other, for example, in the number of coded bits and a favorable one of them i~ selected as a coded output. Let ~one-dimensional]
and ~two-dimensional] represent the numbers of coded bits obtained by coding a coding line by the one-dimensional and the two-dimensional coding system, respectively When [one-dimensional~ > [two-dimensional], the two-dimensional coding is used as a result of a judgement that the amount of information by the one-dimensional coding is larger than that by the two-dimensional coding, whereas when [one-dimensional] _ [two-dimensional], ln the one dimensional coding is employed for the line to be coded as a result of a judgement tllat the amount of information by the one-dimensional coding is smaller than that by the two-dimensional coding.
In accordance with the present invention, there is provided a transmission method for a facsimile signal, in which a two-level facsimile signal obtained by scanning an original picture and successively ..
sampling the scanning output into picture elements is received, as an input, and in which the position of an information change picture element having changed from one to the other of two signal levels is coded and sent out, the improvement of the method comprising: a first step of setting a 20. starting picture element on a coding scanning line to be coded from which tlle coding starts; a second step of detecting a first information change picture element lying next to the starting picture element on the coding scanning line; a third step of detecting a first reference picture element, which is a first information change picture element lying after a picture element just above the starting picture element on a reference scanning line immediately preceding the coding scanning line and has a signal level different from that of the starting picture element, and a second reference picture element of an information change picture element next to the first reference picture element; a fourth step of detecting, 3n as a first mode, the state in which the second reference picture element ~_ LS

precedes a picture element just above the first information change picture element b~ more than n ~n being O or a positive integer) picture elements; a fifth step of detecting, as not the first mode, the state in which the second reference picture element does not precede a picture element just above the first information change picture element by more than n picture elements; a sixth step of comparing a first correlation between the starting picture element and the first information change picture element with a second correlation between the first information change picture element and the first reference picture element when the abovesaid 10 state is detected as not the first mode; a seventh step of coding the presence of the first and second reference picture elements as the first mode and setting the picture element just below the second reference picture element as the starting picture element in the first step when the first mode is det0cted; an eighth step of coding a distance between the starting picture element and the first information change picture element as a second mode and setting the first information change picture element as the starting picture element in the first step when the first correlation is higher than the second correlation; a ninth step of coding the distance between the first information change picture element and the first ref-2n erence picture element as a third mode and setting the first information :~
change picture element as the starting picture element in the first step when the first correlation is not higher than the second correlation; and a tenth step of sending out the coded outputs of the seventh, eighth and ninth steps after combining them into a composite signal of two-dimensional codes.
This invention will be described in details hereinafter with reference to the accompanying drawings, in which:
Figures 1, 2, 3A, 3B, 6, 7, 8A, 8B, 8C, 11 and 16 show examples of facsimile signals, explanatory of the principles of this 3a invention;

~ -4a-~L2~4~

Figure 4A illustrates in block form an embodiment of this invention;
Figures 4B, 4C and 4D illustrate in block forrn specific operative examples of circuits for use in the embodiment of Figure 4A;
Figure 5A shows in block form an example of a decoding device for a facsimile signal encoded b~ the embodiment of Figure 4A;
Figures 5B, 5C and 5D show in block form specific opera-tive examples of circuits for use in the decoding device of Figure 5A;
Figure 9 shows in block form another embodiment of this 10 ..

2~ :.

'-~ tF~
.~ -4b-E;4L5 i nven tion Fig. lOA illustrates in block form an example of a de-coding device for a facsimile signal encoded by the embodiment of Fig. 9 ;
Fig. lOB illustrates in block form a specific operative example of a circuit for u~e in the decoding device of E~ig. lOA;
Fig. 12 shows in block form another embodim~nt of this invention ;
Fig. 13 shows in block form an example of a decoding device for a facsimile signal encoded by th~ embodiment of Fig. 12 ;
Fiqs. 14 and 17 are ~ock diagrams each il~ustrating- another embodiment of this in~ention ; and Fig. 15 is a block diagram illustrating an example of a decoding device for a facsimile signal encoded by the embodiment of Fig. 14.
A detailed description will be given of specific operative examples of this invention.
Figs. 1, 2, 3~ and 3B illustrate examples of acsimile sig-nals, blank blocks representing white picture elements and hatched blocks black picture elements.
At first, a coding start picture element aO and other change picture elements are defined as follows :
aO : a starting picture element on the coding line Lc with which the coding starts along the scanning direction SD ;
a~ : a change picture element nex~ t~ aO on the coding line ;
bl : a first change picture element on the reference line Lr occurring after the picture element just above aO and having a binary signal value different from that of aO ;

~- 5 --' ' ';

~2i~645i ~ 2 ~ a change picture element next to bl on the refer-ence line.
As will hereinbelow be described, the picture elements on the coding line and the reference line are succ~ssively oollat-ed wi,h each other to detect the change picture elements onthe both scanning lines for coding~
(Procedure 1) : In a case where the two change picture ele-ments bl and b2 on the reference line are detected prior to the change picture element al on the coding line (refer to Fig. 2), thi~s state is recoqni ed as "a Pass mode" and ~he chanqe picture elements bl and b2 are coded with a Pass mode code, for example, " 1110" trefer to the column of the Pass mode in Table 1), by which a starting picture element for the next coding is set at a picture element a'0 on the coding line just under the picture element b2.
(ProceduIe 2) : In a case where the change picture element al is detected on the coding line prior to the change picture element bl on the reference line ~refer to Figs. 3A and 3B), scanning of the picture elements proceeds until the change pic-ture element bl occurs ~nd the number of coded bits ~aOal~ isobtained by adding the coded bits of a distance aOal to the bits of a Horizontal mode code "lllln. At the same tim~, the number of coded bits ~blal] for co~ing a di~tance b~al as a Vertical mode is obtained (re~er- to Table 1).

~28~5 Table 1 _ Mode to be coded Code . _ Pass mode blb2 1110 Horizontal mode aOa~ 1111 + MH(aOa~) blal = 0 0 b~al = ~1 100 Vertical mode blal = -1 101 blal _ 2 1100 ~ D(bla~ 1) blal _ -2 1101 + D(¦blal¦ -1) 10 xy xy: white Y _ . n D(n)

3 1000 10 4 00~1

4 1011 011 5 00001 In the column of the "Vertical mode" in Table 1, "~"
indicates the case of the picture element al being detected before the picture element bl and n +1l the case of the picture element al being detected after the picture element bl. These coded bit numbers are compared with each other to select any one of coding modes in accordance with the following conditions:
a) [aOal] > [blal]
In a case where this condition is established ! it is judged that a high correlation exists betwe~n the chanqe picture ele-ment al to be coded and the reference pictuxe elem~nt bl and . .
. ~, , `' . ~

the distance blal is selected as the ~ertical mcde to shift a new starting pic-ture element to the position of the picture element a,.
For example, in the case of Fig. 3A, Lblal~ = "110101" =
6 bits and [aOal] = "11111000" = 8 bits ; consequently, the condition [aOal] _ [blal] is established. Then, the picture element al is encoded by a Vertical mode so that the coded signal "110101" is generated.
b) [aOal] _ [blal]
When this condition is set up, it is judged that a high correlation exists between the change picture element al to be coded and the starting picture element aO and coding of the distance aOal is achieved following the Horizontal mode code "1111", shifting a new starting picture element to the position of the picture element al.
For example, in the case ~f Fig. 3B, [bla~ 6 =
"110100001" = 9 bits and [aOal] - nlllllOOO~ = 8 bits ; con-se~uently, the condition [aOal] _ [blal] is ~stablished and the coded output of the picture element al becomes "lllllOOO"o In the above description, the expressions (a) and (b) are mentioned as the conditions for seleoting either the Horizontal mode or the Vertical mode bu~ other conditional expressions can be used, such as follows :
(a) : [aOal] > [blal] -~ m (~ beinq an inteqer) (b) : [aOal] _ [b~al] + m (m being an integer) Alternatively, if use is made of the distances aoal and bla before coding, (a) : aOal > blal + m (m being ~n integer) (b) : aOal ~ blal + m (m being an integer) ~3~Z~6~5 Moreover, in the column of codes in Table 1, a MH code (a modified Huffmann code, for particulars, refer to CCITT
Recommendation To 4) and a bit-by-bit code D(n) are used ;
but it is a ma~ter of course that the present invention is not limited specifically to the use of such codes and can be achiev-ed with ordinary variable length codes.
Besides, in the procedure 1l it is conditioned that the change picture elements just above the picture elements aO and a~ are not regarded as b~ and b2 ; but the condition can be modified such that the change picture element just above the picture element aO or al is included in bl and b2, or that the change picture elements are not regarded as bl and b2 unless they are not spaced more than _ (_ being 0 or a positive in-teger) picture elements apart from the picture elements aO and al.
As described in detail above, in the present invention, addresses of change picture elements to be coded are succes-sively coded and, in this case, the addresses are each coded using a relative distance between the change picture element to be coded and a selected one of the change picture elements already coded.
A brief description will be made of an example of boundary conditions which are utilized when this inventi~n is reduced into practice, althou~h it does not define the essence of the invention.
(1) Coding of a starting picture element on each scanning line : .
A change picture element from white to black is always used as a first change picture element on each line to be coded.

. .
: . , ;

Accor~ingly, in a case of the first picture element being black, it is made the first change picture element or the first picture element is compulsorily made white.
Further, the first starting picture element aO on each coding line is set up at the position of the first picture element.
(2) Coding of a terminating picture element on each scan-ning line :
The terminating picture element (in CCITT Recommend-ation T. 4, one line consists of 1728 picture elements) ofeach line is coded on the assumption that a change picture element lies next to it.
The following will describe examples of circuits for carrying this invention into practice in accordance with the principles described above~
Fig. 4A illustrates an example of a coding de~ice.
Reference numeral 1 indica~es an input ~erminal for a sampled two-level facsimile signal ; 2 and 3 designate line memories, each storing signals of one line ; 4 identifies a memory for storing the level of starting picture element aO; 5 denotes an address control circuit for controlling addresses of memories 2 ~nd 3 and for generatins an end of line signal EOL ; 6 re-presents an exclusive O~ circuit ; 11 and 12 show change picture element detectors, each composed of a l-bit memory 420 and an exclusive OR circuit 421, as shown in Fig. 4B ; 21, 23 and 24 refer to detectors for detecting the change picture ele~ents a~, bl and b2, respecti~ely ; 25 indicates a blal direction detector ; 32 and 33 designate count~rs ; 40 identifies a pass mode detector ; 52, 53 and 54 denote coders ; 60 repr2sents a 1:~l28~

comparator for comparing the numbers of coded bits with each other ; 71, 72, 74 and 75 show gates ; 81, 82 and 84 refer to address registers, each foxmed by a counter ; 90 indicates a signal combiner ; and 100 identifies an output terminal.
For the sake of brevity, a memory shift pulse generator, a counter clock pulse generator, etc. are not shown ; but these do not exert influence on an understanding of the essence of the operation of the present invention.
Next, the construction and operation of this embodiment will ~e described in detail. A facsimile signal to be coded is provided from the input terminal 1 ~o the coding line memory 2 for storage therein. Before this time, as a signal of a reference line, a signal of the preceding line stored in the line memory 2 is transferred to the reference line memory 3 for 1~ storage therein. ~he a~ memory 4 has stored ~herein level of the starting picture element aO, as will be described later on.
Reading of the coding line memory 2 and the reference line memory 3 simultaneously starts from the position of the start-ing picture element aO under the can~rol of the address control circuit 5. The change picture element detector 11 compares a picture element signal read out o~ the line memory 2 with an immediately preceding picture element signal, ~uccessively, and as a result it generates an output "0" or ~1" in dependence on whether the former signal is of the same level as the latter signal or not. The change picture element detector 12 detects change picture elements on the line memory 3 successively by the same manner as the detector 11. The bl detector 23 is an AND circuit which provides "1" on an,~output line b1p when a change picture element is detectad by the change picture element ~86~

detector 12 and the detected ch~nge picture element level differs from that of the starting picture element aO, that is, when the output from the exclusive OR circuit 6 is "1". The b2 detector 24 provides "1" on an output line b2p in a case where a change picture element is detected by the change picture element detector 12 after detection of the change picture ele-ment bl by the bl detector 23 ; this bl detector 24 can be made up of one flip-flop and an AND circuit. The Pass mode detector 40 is an AND circuit which pro~ides "1" on an output line p, judging that the mode of operation is the Pass made in a case where the picture element a, has not been detected at the moment of occurrence of "1" on the output line b2p (in this case, aln which is the output Q of a flip-10p in the ~1 detec-tor 21 is "1"), as will be described later. With "1" on the output line ~, the Pass mode coding circuit 54 yields a Pass mode code "1110", which is applied to the signal combiner 90.
Following thi~, a new starting picture elemen~ aO is shifted to the position just below the picture element b2 in the follow-ing manner : Upon occurrence o N 1" on the line b2p, the b2 address register ~1 stops counting of pulses from the address control circuit 5 and stores this status. This information is applied via the gate 74 to the aO address register 84 for addition to its content when the Pass mode detector 40 produces "1" on the line p. Besides, the al address register 82 stops counting of pulses from the address control circuit 5 upon occurrence of "1" on the line alp and this information is pro-vided via the gate 75 to the aO address register 84 for addi-tion to its content when the comparator 60 produces "1" on a line V or h. The contents of the aO address register 84 are applied to the address control circuit 5 to re-start the cod ing operation with the new starting picture element. The ad-dress con~rol circuit 5 has such a construction as shown in Fig. 4D, which stores the contents of the aO address register 84 in a register of a memory drive circuit 430 and increases a memory read-out address one by one upon each occurrence of a pulse from a pulse generator 431 to read information of the line memories 2 and 3 simultaneously bit by bit from the a0 address in the register of the memory drive circuit 430.
Further, upon each reception of the contents of the aO address register 84, the address control circuit applies the new start-ing picture element level to the a0 memory 4 via the coding line memory 2. The contents of the memory drive circuit 430 are compared in a comparator 432 with contents of an address memory 433 of the end picture element of one line to generate an end of line signal EOL.
The first change picture element detector 11, when detect ing a change picture element, provides an output "1" to the al detector 21 (a flip-flop3. As a result of this, the information on the lines alp and aln change ~rom ll0" to "1" and from "1"
to "0", respectively The a0al counter 32 starts counting of pulses from the moment of setting a~ in the address control circuit 5, and stops the counting upon reception of "1" from the line alp and provides the count value to the aOal coding circuit 52O The a~al coding circuit 52 encodes the count value with "1111" added to its head, using a cod~ table such as shown in the column oX the ~orizontal mode of Table 1. The b~al counter 33 receives the outputs from the lines blp and alp so that it starts pulse counting with a ~irst appearing "1"

,, ~, : , :

~2~

in either one of the lines blp and alp and stops the counting with a next appearing "1" in the other. To the blal direction detector 25 are also supplied the outputs from the lines blp and alp and, with the circuit construction shown in Fig. 4C, comprising the AND circuits 423 and 424 and two flip-flop circuits 425 and 426, this detector outputs "1" on a line ~
when "1" of the line blp appears earlier than or simultaneously with "1" of the line al but, in the opposite case, provides an output "1" on a line ~.
The blal coding circuit ~3 encodes b~a~ with a sign + or - added thereto on the basis of the count value of the bla counter 33 and the output of the line ~ or - ~rom the b~al direction detector 25, as shown in the column of the Vertical mode of Table 1. The bit numbers encoded by the c~ding circuits 52 and 53 are compared in magnitude with each other in the comparator 60 ; when the condition [aOa~] _ [b~a,] is establish~
ed, "1" is provided on the line V tthe Vertical mode), whereas when this condition is not established, "1" is provided on the line h (Horizontal mode). In a case of the Vertical mode in which "1" is outputted on the line V of the comparator 60, the coded signal of the blal coding circuit 53 is provided via the gate 71 to the signal combiner 30. On the othex hand, in the Horizontal mode in which "1" is yielded on the line h, the gate 72 is opened to apply therethrough the coded signal of the a0al coding circu.it 52 to the signal combiner 90. The sig-nal combiner 90 combines the coded signals applied thareto from the Pass mode coding circuit 5~ and the gates 71 and 72 into a composite signal, which is provided on the output line 100 after being converted into an output signal train~

~2~ 5 For the sake of brevity, the conditions for resetting the detectors, registers, coun~ers and so forth are neither described in the foregoing nor shown in the drawings ; but, required ones of these circuits (the b2 detector 24, the al detector 21, ~he registers 81 and 82, the bla~ direction detec-tor 25, the counters 32 and 33 and so forth) are reset for each setting of the picture element aO.
The interruption of the operation of this coding devlce is placed under the control of the address control circuit 5.
Namely, the aO address is always watched by the address con-trol circuit 5, the coding is stopped at the moment when the aO address becomes a line terminating picture element and the aO address is newly set to a line starting picture element and then coding of the subsequent line is resumed.
The above is the operation of the coding device of Fig.
4A and decoding is achieved by reversing the abovesaid steps.
An example of a decoding device is shown in Fig. 5A. Reference numeral 201 indicates an input terminal ; 202 designates an input buffer memory ; 203 identifies a mode code identify cir-cuit ; 211 and 212 denote line memories ; 213 represents an aO
memory ; 221 and 222 show address control circuits ; 231 and 232 refer to decoding circuits ; 240 indicates a change pic-ture element detector ; 251 and 252 designate a bl detector and a b2 detector, respectively ; 261 and 262 identify an adder and a subtractor, respectively ; 271 and 272 denote counters ; 281 to 285 represent gates ; 291, 292 and 294 show OR circuits ; 293 refers to an e~cl~sive OR circuit ; 300 indi-cates an aO register ; and 310 designates an output terminal.
The following will describe the construction and the :

operation of the decoding device of Fig. 5A in detail. A
coded signal from the input ter~inal 201 is once stored in the input buffPr memory 202. The mode code identify circuit 203 has such a construction as shown ln Fig. 5B, comprising ~esistors 441, 442, 443, 444, 445, 446 and 447 and coincidence circu.its 451, 452, 453 and 454, in which a signal (four bits at most, as shown in Table 1) necessary for mode identifica-tion is read out of ~he input buffer memory 202 to identify the modes of operation, i.e. the Pass mode, the ~orizontal mode and the Vertical mode. When the signal is "1110", it is regarded as indicating the Pass mode and "1" is outputted on a line ~ ; when the signal is nllll~ ~ it is regarded as indi-cating the Horizontal mode and 'tl" iS provided on a line h ;
when the signal is "0", I'lOO or "1100", it is regarded as indicating that the direction of the distance ~a, is plus in the Vertical mode and 1l ln is produced on a line V~ ; and when the signal is "101" or "1101", it is regarded as indicating that the direction of the distance b~a~ is minus in the Verti-cal mode and "1" is yielded on a line V~. The address control circuit 221 has such a construction as depi~ted in Fig. 5C, in which when any one of the outputs ~, V- and V+ from th~ mode code identify circuit is ~lU, pulses are applied to the memory 211 to shift it ~it by bit from the aO address provided from SaO.
When the identify circuit 203 proviee~ "1" on the line (the Pass mode), the address control circuit 221 reads the reference line memory 211 from the a~d~ess of the picture ele-ment aO to start detection of the di~t~nce b~b2. The reference ~Z~6~

line memory has stored thereln inform~tion of the previous line via the coding line memory 212.
The change picture element detector 240 has the const-ruction shown in Fig. 4B and provides an output n 1~ upon each detection of a picture element, whose level is different from the immediately preceding one in the signal train applied from the line memory 211. At the moment when the change picture element detector 240 provides the output "1~, if the detected picture element is different in level from the pic~ure element aO, the output "1" is applied via the exclusive OR circuit 293 to the b~ detector (an AND circuit) 251 to produce an output "1" on a line bl . ~he aOb~ counter 272 receives pulses from the address control circuit 221 and coun~s the number of pulses occurring in the time interval from the aO address to bl (until "1" is provided on the line blp). The b2 detec~or 252 outputs "1" on a line b2p when another change picture element is detec-ted by the change picture element datector 240 after detection of the picture element bl. This bl detector c~mprises a flip-flop and an AND circuit. The a~b2 counter 271 receiYeS pulses from the address control circuit 221 and counts them occurring in the time interval from the a~ address to b2 (until "1" is provided on the line b2p) The contents of the aOb2 counter 271 are applied to the aO register 300 via the gate 231, which is opened by the provision of the output "1" on the line ~ of the mode code identify circuit 203. The contents ~f the aO
register 300 are added to the address control circuits 221 and 222, so that the aO address is newly set and the decoding operation is resumed~
In a case where the identify circuit 203 provid~s "1" on 8~

the line V~ or Y- (Vertical mode), the output "1" from the OR
circuit 291 is applied to the address control circuit 221 and the blal decoding circuit 231. As a consequence, decoding relating to the abovesaid bl takes place and the count value

5 of the aob, counter 272 indicates the address of ~h~ picture element bl relative to the picture element aO. The blal de-coding circuit 231 reads signals of one word from the input buffer memory 202 and decodes them. The decoded value is added by the adder 261 to the value of the aubl counter 272 and, at 10 the same time, sub~racted hy the subtractor 262 from the value of the a~bl counter 272. In a case where the output line V+
of the mode code identify circuit 203 is "1", the gate 284 is opened, so that the information of the adder 261 is provided via the OR circuit 292 to the address control circuit 222 and 15 to the aO register 300 via the gate 282~ In contrast thereto if the output line V- of thf~ mode code identify circuit 203 is "1'1, the gate 285 is opened, passing on the information of the subtractor 262 to the address control circuit 222 via the OR circuit 292 and to the aO register 300 via the gate 282.
20 The address control circuit 222 has such a constructiQn as depicted in Fig. 5D, which sets up the address o:E the picture element aO on the basis of the information transmitted thereto via the OR circuit 292, reproduces the picture slement signals on the coding line as the same le~rel as the picture element a0 25 from the picture element aO to a picture element immediately preceding a" and inverts the level of the picture element a relative to the information of the picture element aO. The contents of the aO register 30Q are applied to the address control circuits 221 and 222, newly setting the address of the , ~

~,~lZ~ 5 picture element aO and r~suming decoding.
In a case where the line h of the mode code identify cir-cuit 203 becom~s "1" (Horizontal mode), the a~al decoding cir-cuit 232 reads signals of one word from the input buffer memory 202 and decodes them. The decoded value is added to the address control circuit 222 and the aO register 300 via the gate 283.
The address control circuit 222 sets up the addres of ~he picture element al, reproduces the picture element signal on the coding line as the same level as the picture element aO
from the picture element aO to a picture element immediately preceding al, and makès the level of the picture element al to be different from the level of the picture elemen~ aO. The aO
address register 300 restores ~he address of the picture ele-ment al, so that the a.l. addres~ becomes a new aO addre~s.
This new addresæ is provided to the address cont~ol circuits 221 and 222 to set the aq address and r~-st~rt decoding.
Also in respect of the abo~e d~coding device, the reset-ting conditions for the detectors, the registers, the counters and so forth have been neither described nor shown in the drawings ; but the mode code identify circuit 203, the b2 detector 252, the address control circuits 221 and 222, the counters 271 and 272, the decoding circuits 231 and 232, etc.
are reset for each new setting of the aO addre~s~ The termi-nation of one line is achieved by supervisIng the aO address with the addre~s control circuit 222 and at the ~ment of the address of ~he picture element becoming the ad~res~ of the last picture element of a scanning line, decQding of that line is completed and decoding of the next line is resumed.
In the above embodiment, when the Horizontal mode is .. :..................... , :

~8~9~5 identified, a distance between the starting picture elemen~
and the coding change point is encoded and the code "1111"
indicating the Horizontal mode is added to the encoded value.
For further enhancement of the compression efficiency, however, it is considered that in~the case of the Horizontal mode being identified, the following change picture elements are encoded together and added wi~h one Horizontal mode code "1111".
That is, one Horizontal mode code i5 shared by two change pic-ture elements ; consequently, the compression efficiency is improved. This will hereinbelow be described in detailO
Fig. 6 illustrates examples of facsimile signals, blank blocks representing white picture elements and hatched blocks black picture elements. At first, a coding start pic~ure ele-ment aO and other change picture elemenks are defined as follows :
aO : a starting picture element on the coding line with which the coding starts ;
al : a change picture element next to a~ o~ the coding line ;
a2 : a change picture element next to al on the coding line ;
b~ : a first change picture element on khe reference line occurring after the picture element just above a~ and having a binary signal value different from that of a~ ;
b2 : a change picture element next to b1 o~ the reer-ence line.
As will hereinbelow be describ'ed, the picture elements on the coding line and the reference linç are successi~ly collated with each other to dekect the change picture elements ~ 20 -~286~S

on the both scan lines for coding.
(Procedure 1) : In a case where the two change picture elements bl and b2 on the re~erence line are detected prior to the change picture element a, on the coding line (refer to Fig.7), this state is recoynized as a Pass mode and the change picture elements bl and b2 are cod~d with a Pass mode code, for example, "1110" (refer to the column of the Pass mode in Table 2), by which a starting picture element Eor ~he next coding is set at a picture element alo on the coding line just under the picture element b2.
(Procedure 2) : In a case where the change picture ele-ment al is detected on ~he coding line prior to the change picture element bl on the reference line (refer to Figs. 8A, 8B), scanning of the picture elements proceeds until the change picture element b, occurs and the n~mber of coded bits [aOal]
is ob~ained by adding the coded bi~s of a distance a~a~ to the bits of a Horizontal mode code l'llll". At the same time, the number of coded bits [blal] for coding a distance blal as a Vertical mode is obtained (refer to Table 2).

~o Table 2 Mode Elements to be Code _ Pass mode blb2 1110 _ Hori20ntal mode a~al, ala2 1111 ~ MH(aOa~) ~ MH(ala2) _ blal = O
blal = ~1 100 Vertical mode bla~ = -1 101 bla~ _ 2 1100 + D(blal - 1) :~

bla~ < -2 1101 + D(¦bla , . . ..
. ,. :
~, , ~2~45 ~ xy . white MH(xy) n ~

4 1011 I ~ 00001 The sign~ "-" and "+" are the same as in Table 1. These coded bit num~ers are compared to select any one of coding modes in accordance with the following conditions -a) [aOal] > [blal]
When this condition is established, it is judged that a high correlation exists between the change picture element a, to be coded and the reference picture element bl, and the value of the distance blal coded in the Vertical mode is selected to shift a new starting picture element to the position of the picture element al.
For exa~ple, in the case of Fig. 8A, [b~al] = "110101" =

6 bits and [aOal] = "11111000" = 8 bit~ ; consequently, the conditio~ [aOal] ~ [blal] is established. Then th~ picture element al is encoded by a Vertical mode so that the coded sig-nal "110101" is generated.
b) [aOal] _ [blal]
When this condition is set up, it is judged that a high correlation exists between the change picture element al to be coded and the starting picture element aO, and it is decided to perform coding in the ~orizontal mode until a change picture element a2 appears after al ; thus, collation proceeds until the change picture element a2 occurs and code generation of the distances aOal and ala2 is achieved following the gener-ation of Horizontal mode code, for example, "1111", thereby shifting a new starting picture element to the position of the picture element a2.
For example, in the case of Fig. 8B, [blal] = -6 =
"110100001" = 9 bits and [aOal] = "11111000" = 8 bits ; con-sequently, the condition [aDal] < [b~all is established and the coded outputs of the picture elements al and a2 become 10 "11111000" and "011".
In the above description, the expressions (a) and (b) are mentioned as the conditions for selecting either the Horizontal mode or the Vertical mode but other conditional expressions can be used, such as follows :
(a) : [aOal] > [b~al] + m (m being an integer) (b) : [aOal] _ [blal] ~ m (m being an integer) Alternatively, if use is made ~f the distances aOal and b~a before coding, (a) : aOal > bla, ~ m (m being an integer) (b) : aOal _ b~a~ + m (m being an integer) Moreover, in the column of codes in Table 2, a MH code (a modified Huffmann code, for particulars, refer to CCITT
Recommendation T.4) and a bit-by-bit code D(n) are used ; but it is a matter of course that the present invention is not limited specifically to the use of such codes and can be achieved with ordinary variable lPngth codes.
Besides, in the proceduxe 1, it is conditioned that the change picture elements just above the picture elements aO and al are not regarded as bl and b2 ; but the condition can be 36~S

modified such that the change picture element just above the picture element a~ or al is included in b1 and b2, or that the change picture elements are not regarded as bl and b2 unless they are not spaced more than n (n being 0 or a positive in-teger) picture elements apart from the plcture elements aO andal .
As described in detail above, in the present invention, addresses of change picture elements to be coded are succes-sively coded and, in this case, the addresses are each coded using a relative distance between the change picture element to be coded and a selected one of the change picture elements al-ready coded. Where this selected change picture element is the starting picture element aO on the codiny line, the address of the next change picture element a 2 to be coded is also coded using the relative distance between it and the picture element aO. As a consequence, the change picture elements whose ad-dresses are coded using the distances between them and the starting picture elements on the coding line, are always in pairs. In a case of using the relative distance between change picture elements to be coded and a change picture element on the immediately preceding re~erence line, the chang~ picture elements on the coding line are coded individually.
A brief description will be made of an example of boundary .
conditions which are utili2ed when this invention is reduced into practice, althrough it does not define the essence of the invention.
(1) Coding of a starting picture element on each scan-ning linP -A change picture element from white to black is always - 2~ -36~5 used as a first change picture element on each line to be coded. Accordingly, in a case of the first picture element being black, it is made the ~irst change pic~ure element or the first picture element is compulsorily made white.
Further, the first starting picture element aO on each coding line is set up at the position of the first picture element.
~2) Coding of a terminating picture element on each scan-ning line :
The terminating picture element (In CCITT Recommendation To 4. one line consists of 1728 picture elements ; accordingly, the terminating picture element is the 1728th picture element.) of each line is coded on the assumption that a change picture element lies next to it.
The following will describe examples of circuits for carry-ing this invention into practice in accordance with the princi-ples described above.
Fig. 9 illustrates an example of a codinq devica. Refer-ence numeral 1 indicates an input ~erminal for a sampled two-level facsimile signal ; 2 and 3 designate line memories, each storing signals of one line ; 4 identifies a memory fox stor-ing the level of starting picture elemen~ aO ; 5 denotes an address control circuit for controlling addresses of memories 2 and 3 ; 6 represents an exclusive OR circuit ; 11 and 12 show change picture element detectors, each compose~ of a l-bit memory and an exclusive OR circuit, as shown in Fig. 4B ; 21, 22, 23 and 24 refer to detectoxs for detecting the change picture elements al, a2, b~ and b2, respectively ; 25 indicates a blal direction detector ; 31, 32 and 33 designate counters ;

86~5 40 identifies a Pass mode detector ; 51, 52, 53 and 54 denote coders ; 60 represents a comparator for comparinq the numbers of coded bits with each other ; 71, 72, 73, 74, 75 and 76 show qates ; 81, 82, 83 and 84 refer to address registers ; 90 indi-cates a signal combiner ; and 100 identifies an output termi-nal. For the sake of brevity, a memory shift pulse generator, a counter clock pulse genera~or, etc. are not shown ; but these do not exert influence on an understanding of the essence of the operation of the present invention.
Next, the construction and operation of this embodiment will be described in detail. A facsimile signal to be coded is provided from the input terminal 1 to the coding line memory 2 for storage therein. Before this time, as a signal of a reference line, a signal of the preceding line stored in the line memory 2 is transferred to the reference line memory 3 for storage therein. The a0 memory 4 has stored therein level of the starting picture element aO, as will be described la~er on.
Reading of the coding line memory 2 and the reference line memory 3 simultaneously starts from the position of the starting picture element aO under the control of the address contrsl circuit 5. The change picture element detectors 11 and 12 each comprise an exclusive OR circuit and a l-bit memory, as shown in Fig. 4B, and compare the picture element ~ignals read out of the each line memories 2 and 3 with immediately pre~eding picture element signals to output "0" or "1^' in dependence on whether the former signals are of the s~me level as the latter signals or not, respectively. The bl detector 23 is an AND
circuit which provides IllU on an outp~t line blp when a change picture element is detected by the ~econd change picture element .,j .

~Z8~

detector 12 and the dete~ted change picture element level differs fro~ tha~ of the s~arting picture element aO, that is, when the output from the exclusive OR circuit 6 is "1". The b2 detector 24 provides "1" on an outpu~ line b2p in a case where a change picture element is detected by the change pic-ture element detector 1~ after de~ection of the change plcture element bl by the bl detector 23 ; this b2 detector 24 can be made up of one flip~flop and an AND circuit. The Pass mode detector 40 is an AND circuit which provides "1" on an output line p, judging that the mode of operation is the Pass mode in a case where the picture element a, has not been detected at the moment of occurrence of "1" on the output line b2p (in this case, aln which is the output Q of a flip-flop in the a~
detector 21 i5 ~ as will be described later. With "1" on the output line ~, the Pass mode coding circuit 54 yields a Pass mode code "1110", which is applied to the signal combiner 90. Following this, a new starting picture element (a0) is shifted to the position just under the picture element b2 in the following manner : Upon occurrence of "1" on the line blp, the b2 address register 81 stops counting of pulses from the address control circuit 5 and stores the count value.
These contents are applied via the gate 74 to the aO address register 84 when the Pass mode detector 40 produces "1" on the line ~. The contents of the aO address register 84 are applied to the address control circuit 5 ~o re-start the coding oper-ation with the new starting picture element~
The first change picture elemen~ detector 11, when detect-ing a change picture element, provides an output "1" to the a~ ;
detec~or 21 (a flip-flop~. As a result of this, the information on the lines alp and aln change from "0" to "l" and ~rom "1"
to "0", respectivley. The a 2 detectox 22 is a flip-flop which produces "l" on a line a2 when a change picture element is detected by the change picture element detector 11 after the picture element a1 is detected by the al detector 21 ("l" on the line alp). The aOa~ counter 32 starts counting of pulses from the moment of setting a~ in the address control circuit 5, and stops the counting upon reception of "l" from the line a1p and provides the count value to the aOa~ coding circuit 52. The aOal coding circuit encodes the count value with "llll" added to its head, using, for example, ~uch a code table as shown in the column of the Horizontal mode of Table l. The a~a2 counter 31 starts counting with "l" on the line alp and stops the counting with "l" on the line a2p and provides the count value to the ala2 coding cixcuit 51.
The ala2 coding circuit 51 encodes the count value using such a code table, for example, as shown in the column MH(xy) of Table 2. The b~al counter 33 receives the outputs from the lines blp and alp ~o that it starts pulse counting with a first appearing "l" in either one of the outputs blp and alp and stops the counting with a next appearing "ll' in the other. To the bla~ direction detector 25 are also applied the outputs from the lines blp and alp and, with th~ circuit construction shawn in Fig. 4C, this detector o~tputs "l" on a line + when "1" of the lin~ blp appears earlier than or simultaneously with "1" of the line alp but, in the opposite case, provides an output "l n on a line -.
The blal coding circuit 53 encodes bla, with a sign + or - added thereto on the basis o~ the count value of ~he bla s counter 33 and the output of ~he line ~ or - from the b,al direction detector 25, as shown in the column of the ~ertical mode of Table l. The bit num~ers encoded by the coding cir-cuits 52 and 53 are compared in magnitude with each other in the comparator 60 ; when ~he conditlon ~a0al] > [blal] is established "1" is provided on the line V (Vertical mode), whereas when this condition is no~ established, "1" is pro-vided on the line _ (Horizontal mode). In a case of the Vert-cal mode in which "l" is outputted on the line V oP ~he com-parator 60, the coded signal of the bla~ coding circu~t 53 isprovided via the gate 71 to the signal combiner 90. On the other hand, in the ~orizontal mode in which "l" iæ yielded on the line h, the gates 72 and 73 are opened to apply therethrough the coded signals of the a0al coding circuit 52 and the ala2 coding circuit 51 to the signal combi~er 90. The signal com~
biner 90 combines the coded signals applied thereto from the Pass mode coding circuit 54 and the gates 71, 72 and 73 into a composite signal, which is provided on the output line 100 after being converted into an output signal train.
For the sake of brevity, the conditions ox resetting the detectors, registers, counters and so forth are neither de~
scribed in the foregoing nor shown in ~he drawings ; but, re-quired ones of these circuits (the b2 detector 24, the a detector 21, the a2 detector 22, the register~ 81, 82 and 83, the blal direction detector 25, the counter~ 31, 32 and 33 and 50 forth) ar~ reset for each setting of the picture element a0.
The interruption of the operation of this coding device is placed under the control of the address control circuit.
Namely, the a0 address is always watched by the address control circuit 5, the coding is stopped at the moment when the a~
address becomes a line terminating picture element and the aO address is newly set to a line starting picture element and then coding of the subsequent line is resumed.
The above is the operation of the coding device of Fig. 9 and decoding is achieved by reversing the abovesaid steps.
An example of a decoding device is shown in Fig. lOA. Refer-ence numeral 201 indicates an input terminal ; 202 designates an input buffer memory ; 203 identifies a mode code identlfy 10 circuit ; 211 and 212 denotP line memories ; 213 represents an aO memory ; 221 and 222 show address control circuits ; 231, 232 and 233 refer to decoding circuits ; 240 indicates a change picture element detector ; 251 and 252 designate a b~ detector and a bz detector, respectively ; 261 and 262 identify an adder 15 and a subtractor, respectively ; 271 and 272 denote counters;
281, 282, 283, 284, 285 and 286 represent gates ; 291, 292 and 294 show OR circuits ; 293 refers to an exclusive OR circuit ;
300 indicates an aO register ; and 310 ~esignates an output terminal.
The following will describe the construction and the oper-ation of the decoding circuit of Fig~ lOA in detail. A coded signal from the input terminal 201 is once stored in the in-put buffer memory 202. The mode code identify circuit 203 has such a construction as shown in Fig. 5B, in which a signal (for example, four bits at most, as shown in Table 2) necessary for mode identification is read out of the input buffer memory 202 to identify the modes of operation, i.e. the Pass mode, the Horizontal mode and the Vertical mode. When the signal is "1110", it is regarded as indicating the Pass mode and "1"

.

is outputted on a line ~ ; when the ~ignal is "1111", it is regarded as indicating the Horizontal mode and "1" and is provided on a line h ; when the signal is "0", ~loa~ or "1100", it is regarded as indicating that the direction of the distance blal is plus in the Ver~ical mode and "1" is produced on a line V+ ; and when the signal is "101" or "1101", it is regard-ed as indicating that the direction of the distance blal is minus in the Vertical mode and "1" is yielded on a line V-.
The address control circuit 221 has such a construction as depicted in Fig. 5C, in which when any one of the outputs p, V- and V+ from the mode code identify circuit is "1", pulses are applièd to the mèmory 211 to shift it bit by bit from the a0 address provided from Sa0.
When the identify circuit 203 provides "1" on the line p (the Pass mode?, the address control circuit 221 shifts the reference line memory 211 from the address of the picture ele-ment aO to start detection of the distance b~b2. The reference line memory has stored therein information of the previous line via the decoding line memory 212. The change picture ele-ment detector 240 has the construction shown in Fig. 4B andprovides an output "1" upon each detection of a picture element different from the immediately preceding one in the signal train applied from the line memory 211. At the moment when the change picture element detector 240 provides the output "1l', if the detected picture element is different in level from the picture element aO, the output "1" is applied via the exclusive OR
circuit 293 to the bl detector (an AND circuit) 251 to produce an output "1l' on a line blp. The a0bl counter 272 receives pulses from the address control circuit 221 and counts the 69L~i number of pulses occurring in the time interval from the aO
address to b~ til "1" is provided on the line blp). The b2 detector 25~ outputs "li~ or. a line b2p when ano-~er change picture element is detected by the change pic~ure element detector 240 after detection of the picture element b, ("1"
on the line blp). This bl detector comprises a flip-flop and an AND circuit. The a ob~ counter 271 receives pulses from the address control circuit ~21 and counts them occurring in the -time interval from the aO address to b2 (until "1" is pro-vided on the line b2 ) Upon occurrence of "1" on the Lineb2p, the address con-trol circuit 221 once stops sending out of shift pulses. The information of the a~b2 counter 271 is applied to the aO register 300 via the gate 281, which is opened by the provision of the output "1" on the line p of the mode code identify circuit 203. The information of the aO
register 300 is added to the address control circuits 221 and 222, so that the aO address is newly set and the decoding oper-ation i~s resumed.
In a case where the identify circuit 203 provides "1" on the line V+ or V- (the Vertical mode), the output "1" from the OR circuit 291 is applied to the address control circui-t 221 and the bla~ decoding cixcuit 231. As a consequence, decoding relating to the abovesaid bl and b2 takes place and the count value of the a~bl counter indicates the address of the picture element bl relatiYe to the picture element aO. The b~a, de--coding circuit 231 reads signals of one word from the input buffer memory 202 and decodes them. The decoded value is added by the adder 261 to the value of the aObl counter 272 and, at the same timel subtracted by the subtractor 262 from the value ~ , ~L2~

of the a0bl count~r 27~. In a case where the output line V+
of the mode code ldentify circuit 203 is "1", the yate 284 is open~d, so that the contents of the adder 261 are provided via the OR circuit 292 to the address control circuit 222 and to the aO register 300 via the gate 282. In contract thereto, if the output line V- of the mode code identify circuit 203 is "1", the gate 285 is opened, passing the conkents of the subtractor 262 to the address control circuit 222 via the OR
circuit 292 and to the aO register 300 via the gate 282.
The addxess control circuit 222 has such a construction as depicted in Fig. 10B, which sets up the address of the picture element al on the basis of the information transmitted thereto via the OR circuit 292, reproduces picture element sig-nals on the decoding line memory 212 from the pictuxe element a0 to a picture element immediately preceding al to be the same level as the picture element a0 and inverts the level of the picture element al relative to the level of the picture element aO. The contents of the aO register 300 are applied to the address control circuits 221 and 222, newly setting the address of the picture ele~ent a0 and resuming decoding.
In a case where the line h of the mode code identify cir-cuit 203 becon~es "1" (the Horizontal mode), the aOa~ and ala2 decoding circuits 232 and 233 successively read si~nals of two words from the input buffer memory 202 and the a0al decoding circuit 232 decodes ~he first one word and the ala2 decoding circuit 233 the second one word. The decoded values are added to the address control circuit 222 and the aO register 300 via the gates 283 and 286. The address control circuit 222 sets u~ the addresses of the picture elements al and a0, reproduces ' ~
~' ~ ,. ..

picture element signals on the decoding line memory 212 from the picture element aO to a picture element immediately preceding al to be the same as the level of the picture ele-ment aD and inverts the level of the picture element al and, thereafter, reproduces picture element signals from the picture element al to a pic~ure elemen~ immediately preceding a2 to be the same as the level of the picture element al and makes the information of the picture element a2 to be different from the level of the picture element al. The aO address register 300 restoxes the address of the picture elements al and a2, so that the a2 address becomes a new aO address. This new in-formation is provided to the address control circuits 221 and 222 to set the aO address and restart decoding.
Also in respect of the above decoding device, the reset-ting conditions for the detectors, the registers, the countersand so forth have been neither described nor shown in the drawings ; but required ones of them (the mode code identify circuit 203, the b2 detector 252, the address control circuits 221 and 222, the counters 271 and 272, the decoding circuit 231, 232 and 233, etc.) are reset for each setting of the aO
address. The termination of one line lS achieved by super-vising the aO address with the address control circuit 222 and at the moment of the address of the picture element aO becoming the address of the last picture element of a scanning line, decoding of that line is completed and decoding of the next line is resumed.
Next, a description will be given of a system of suppres-sing degradation of the picture ~uality of the reproduced picture due to a code error which is another object of this .

~2~ 5 invention. In the coding system of ~his invention, a picture signal of the coding line is encoded using picture signal information of a ~eferen~e line immediately preceding the coding line. Accordingly, on the side of the decoder, the picture signal of the coding lin~ is also decoded using the picture signal information already decoded. Thus, since cod-inq and decoding are performed successively using -~he picture siqnal information of scanning lines i~nediately preceding the coding lines respec-tively, if a code error occurs due to the influence of circuit noises and the like to cause incorrect reproduction of picture signals of a certain line, picture signals of the succeeding lines are not reproduced correctly, resulting in markedly degraded picture quality of the repro-duced picture.
Accordingly, it is necessary to detect occurrence of a code error, to suppress degradation of the picture quality of the line, in which the code error has occured and to rapidly recover from the code error state, so that the deterioration of the picture quality due to the code error do not spread to other lines.
According ~o this invention, these objects are achieved in the following manner : On the side of the coding device, a detectable, so-called self-synchroniæed first contxol code is inserted, fro~ a desired position in a code train, at a predetermined period of a picture signal, for example, immedi-ately before the starting of coding of a line No.l evey four lines (K=4) as shown in Fig. 11 ; picture signal information of the line No.1 is encoded (into, for instance, a run-length code RL) by a one-dimensional method I without using picture . -. j . , , ~ . .

Z86~5 signal informatlon of a line immediately preceding the line Nc.l ; scanning lines No.2, NoO3, ... No~K immedlately follow-ing the line No.l are subjected to the two dimensional succes-sive coding II of this invention ; and a second control code, different rrom the first control code for de-~ecting the occurrence of a code error is inserted just before tne coded signal of each line.
On ths side of the decoding device, when the self-synchronized first control code is detected, it is decoded as the line No.l without using information of the immediately preceding line on the assumption that the directly following code train has been encoded into a run-lenyth code, RL. When the second control code is detected, it is decoded using in-formation of the immediately preceding line on the assumption that it has been encoded according to ~his invention. Directly after completion of decoding of each line, the presence or absence of the first or second control code is checked to effect error checking. When an error is detected, the decoded line, in which the error is detected, is subjected to processing such as replacement by a picture signal o-E the immediately preced-ing line to thereby suppress deterioration of the picture quality. Upon detection of the error, the decoding operation is once stopped , but when the self-synchronized first control code is detected, decoding of the run-length code RL is immedi~
ately started to restore from the error state.
Fig. 12 illustrates in block form a coding device embody-ing the present invention based on such principles, and Fig. 13 a corresponding decoding device. A facsimile picture signal input line 1 is connected via a switch 101a to an RL coder 102 ~ 36 $~ 8~4S

every K lines under the control of a switch con~rol circuit 101. At this time, a first control code generator 104 gener-ates a firs~i con~rol code and the RL coder 102 encodes a line (No. 1) into a run-leng~h codeO Upon completion of this encoding, the switch 101a is connected to a two~dimensional coder 103 of this invention ~o achieve two-dimensional coding of lines No. 2 to No. K according to this invention and a second control code is inserted by a second control code in-serting circuit 105 just before the coded signal of each scan-ning line.
On the side of the decoding device shown in Fig. 13, whenthe first control code is detected by a first control code detector 106, the run~length code is decoded by a run-length code decoder 107 for one line (No. 1) only and the reproduced lS picture element information is stored in a line memory 108 and, upon completion of decoding of the line No. 1, the content of the line memory 10B is transferred to a line memory 109.
Thereafter, successive decoding of the lines No. 2, No. 3, ..
.. No. K corresponding to the coding of this invention is effected by such a decoder 110 as shown in Figs. 5A and 10A
using the content of the line memory 109. Upon completion of decoding of each llne, the control codes are detected by the control code detectors 106 and 111, and it is checked by a code error detector 112 for occurrence of a code error.
Once a code error has occurred in a scanning line, no decoding takes place until the scanning line No. K. Then, upon detectlon of the first control code, an ordinary decoding operation is started to restore from the code error state.
As has been described in the foregoing, the present invention has the advantage that highly efficient coding can be achieved without depending on correlation, between adja-cent lines of signals, by properly selecting the two kinds of coding systems in which a signal having high correlation between adjacent lines, such as a two-level facsimile signal, is encoded with high efficiency using a distance between a change picture element ~o be encoded and an adjoining one, and in which in a case of a part having no correlation to a line just above it, just like a first line of a document, a change picture element is encoded using a distance between it and another picture element of the same line.
The present invention has another advantage that by in-serting a self-synchronized first control code, for example, every K scanning lines, encoding only one scanning line into lS run-length codes, encodinq the subsequent scanninq lines accord-inq to this invention and then checkinq for a code error u~on completion of codinq of the said one scanninq line, deqradat-ion of the picture quality due to the code error is prevented from s~readinq, therebY to enable ra~id restoration from the code error state.
In the following, another embodiment of this invention relating to the second object will now be described, in which the two dimensional coding principle as described above and the one dimensional coding principle, such as the run-length coding principle, are adaptively adopted.
Next, an example of the one-dimensional coding will be described. Fig. 8C shows an example of a facsimile signal.
In the one~dimensional coding system, a run from a picture element Cl to a picture element directly before a picture ~Z~6~;

element C2 consists of five black picture elements, and hence is coded into "0011", for example, according to the MH code in Table 1 ; a run from the picture element C 2 to a picture element immediately before a picture element C 3 consists of seven white picture elements, and hence is coded into "1111" ;
and a run from the picture element C3 to a picture element immediately before a picture element C4 consists of two black picture elements, and hence is coded into "11". These coded trains are stored or outputted as a one-dimentional coded line.
The following will describe examples of circuits for applying this invention into practice in accorda~ce with the principles described above.
Fig. 14 is an example of a coding device, in which the part indicated by a dotted enclosure is the same as Fig. 3.
lS A change picture element detector 13 i5 composed of a l-bit memory and an exclusive OR circuit as shown in Fig~ 4B. There are further provided a NAND circuit 7, an AND circuit 8, a counter 34, coders 55 and 56, coded signal memories 91 and 92, a comparator 62, tates 77 and 78, a first control code gener-ator 102, and a second control code generator 101.
Next, the construction and operation of this embodimentwill be described in detail~ A facsimile signal to he coded is provided from the input terminal 1 to the coding line memory 2 for storage therein. ~efore this time, as a signal of a reference line, a signal of the preceding line stored in the line memory 2 is transferred to the reference line memory 3 for storage therein. The aO memory-4 has stored therein level of the starting picture element a0, as will be described later on. Reading of the coding line memory 2 and the reference ~ ~ .

~21~ 5 line memory 3 simultaneously starts from the posi~ion of the starting picture element a0 under the control of the address control circuit S.
The change picture elemen~ detectors 11, 12 and 13 res-pectively are each constructed, as shown in Fig. 4B, andcompare the picture element signals read out of the line memo-ries 2 and 3, respectively, with immediately preceding picture element signals of each line to output "0" or "1" in dependence on whether the former signals are of the same level as the latter signals or not.
The bl detector 23 is an AND circuit which provides "1" on the output line blp when a change picture element is detected by the change picture element detector 12 and level of the detected change picture element differs from that of the start-ing picture element a0, that is, when the output from the ex-clusive OR circuit 6 is "1". The b2 detector 24 procides "1"
on an output line b2p in a case where a change picture element is detected by the,change picture element detector 12 after detection of the change picture element bl by the bl detector ,20 23 ; this bl detector 24 can be made up of one flip-flop and an AND circuit. The Pass mode detector 40 is an AND circuit which provides "1" on an output line ~, judging that the mode ;~
of operation is the Pass mode in a case where the picture ele-ment al has not been detected at the moment of occurrence of "1" on the output line b2p (in this case, a1n which is the output Q of a flip-flop in thé al detector 21 i8 " 1" ), as will be described later. With "1" on the output line p, the Pass mode coder to the coded signal memory 91. Following this, a new starting picture element is shifted to the position just ~ '10 - . . . .

under the picture element b2 in the following manner : Upon occurrence of "1" on the line b2p, the b2 address register 81 stops counding of pulses from the address control circuit 5 and stores the count value. This information is applied via the gate 74 to the aO address register 84 at the moment of the Pass mode detector 40 providlng "1" on the line p. The contents of the aO address register 84 are applied to the address control circuit 5 to re-start the coding operation with the new starting picture element aO~
The change picture element detector 11, ~hen detecting a change picture element, provides an output "1" to the a~ de-tector 21 (a flip-flop). As a result of this, the information on the lines al and a1n change from "0" to "1" and from "1"
to "0", respectively. The a2 detector 22 is a flip-flop which outputs "1" on a line a2p when a change picture element is detected by the change picture element det~ctor 11 after the picture element a~ is detected by the a~ detector 21 ("1" on the line alp~. The aOal counter 32 starts counting of pulses from the moment of setting aO in the address control circuit ~20 5, but stops the counting upon reception of "1" from the line alp and provides the count value to the aOa~ coder 52. The aOa~ coding circuit encodes the count value with "1111" added to its head, using such a code table as shown in the column of the Horizontal mode of Table 1. The ala2 countex 31 starts counting with "1" on the line alp and stops the counting with "1" on the line a2p and provides the count value to the ala2 coder. The a,a2 coder 51 encodes the count value using such a code table as shown in the column MH(xy) of Table 1. The blal counter 33 43ceives the outputs from the lines blp and .
, alp and starts pulse counting with a first appearlng "1" in either one of the outputs and stops the counting wi-th a next appearing "1'l in the other. Tu the blal direction detector 25 are also applied the outputs from ~ lines blp and alp and, with the circuit construction shown in Fig. 4C, this detector outputs "1" on a line + when "1" of the line blp appears earlier than or simultaneously with "1" of the line alp but, in the opposite case, provides an output "1" on a line -.
The blal coder 53 encodes b~al with a sign + or - added thereto on the basis of the count value of the blal counter 33 and the output of the line + or - from the blal direction detector 25, as shown in the column of the ~ertical mode of Table 1. The bit numbers encoded by the coders 52 and 53 are compared in magnitude with each other in the comparator 61 ;
when the condition [aOal~ > [blal] is established, "1l' is pro-vided on the line V (Vertical mode), whereas when this condi-tion is not established, "1" is provided on the line h (Hori-zontal mode). In a case of the Vertical mode in which "1" is outputted on the line V of the comparator 61, the coded signal of the blal coder 53 is provided via the gate 71 to the coded signal memory 91. ~n the other hand, in the ~orizontal mode in which "1" is yielded on the line h, the gates 72 and 73 are opened to apply th~rethrough the coded signals of the aOal and ala2 coders 52 to the coded signal memory 91.
The change picture element detector 13 is a detector for the one-dimensional coding. Upon detection of a change picture element by this detector, the counter 34 starts counting o~
clock pulses Pc and, upon detection of the next change picture 8~4~

element, this counting is once stopped, and -the count value at this moment is coded by the coder S5 or 56 of the nex-t stage.
The output from the counter 34 is coded by the coder 55 .S or 56 in dependence on whether the signal is white or black.
Namely, a signal from the coding line memory 2 and the output from the change picture element detector 13 are applied to the NAND circuit 7 and the AND circuit 8, and the outputs from the NAND circuit 7 and the AND circuit 8 are applied to the coders 55 and 56 respectively ; the coder 55 or 56 operates in depen-dence on whether the outputs from the NAND circuit and the AND
circuit are each "0" (white) or "1" (black). In this manner, the count value of the counter 34 is applied to the coder 55 or 56 and coded therein by the MH code of Table 1, thereafter being provided as a one-dimensional coded train to the coded signal memory 92. The coded output signal thus stored in the coded signal memory 91 is a two-dimensional coded signal, there-as the coded output signal stored in the coded signal memory 92 is a one-dimensional coded signal. These coded signals are applied to the comparator 62 and compared with each other, for example, in the number of bits for each line in the outputs from the memories 91 and 92 for selecting a more advantageous one of the both memory output signals.
Where the one-dimensional coding is judged to be advantage-ous as a result of the comparison in the comparator 62, an out-put Sl becomes "1" to open the gate 78 for passing on the in-formation of the coded signal memory 92 to the signal c~mbiner 110. At the same time, the first control code generator 102 provides a first control code (a first line cynchroni7ing signal .,- : ~

~2~36~5 LSSl), for example, "01111111" representing that the line is a one-dimensional coded line. This control code i5 added to the head of the information of the coded signal memory 92.
In case the two-dimensional coding is judged to be advant-ageous as a result of the comparison in the comparator 62, anoutput S2 becomes "l" to open the gate 77 for applying there-through the information of the coded signal memory 91 to the signal combiner llO. At the same time, the second control code generator lOl provides a second control code (a second line synchronizing signal LSS2), for example, "01111110" indicating that the line is a two-dimensional coded line. This control code is added to the head of the information of the coded signal memory 91. The signal combiner llO combines the control code from the control code generator lO1 or 102 and the signal from the gate 77 or 78 into a composite signal, which is sen~ out from the output terminal 120 after being ~onverted into an out-put signal train.
In a case of producing the first and second control codes in the form of "01111111" and "01111110" respectively, as de-scribed above, in order to make these control codes distingish-able from other codes, it is necessary, for example, to compul sorily insert "0" in the control codes every five "ls" occurring successively in the coded signals, like "11111010" ...".
Needless to say, the decoding side decodes the coded signals removin~ "0" next to "lllll" in the coded signal.
For the sake of brevity, the conditions for resetting the detectors, registers, counters and 90 forth are neither described in the foregoing nor shown in the drawings ; but, required ones o these circuits (the b2 detector 24, the al detector 21, the - 4~

36~5 a2 detector 22, the registers 81, 82 and 83, the b~al direction detector 25, the counters 31, 32 and 33 and so forth) are re-set for each setting of ~he picture element aO.
The interruption of the operation of this coding device is placed under the control of the address control circuit.
Namely, the aO address is always watched by the address control circuit 5, and the coding is stopped at the moment when the aO
address becomes a line terminating picture element, and the aO
address is newly set to a line starting picture element, and then coding of the subsequent line is resumed.
An example of a decoding device for receiving a facsimile signal encoded by the embodiment of Fig. 14 is shown in Fig. 15, in which circuits enclosed by a dotted enclosure are further added to the decoding device shown in Fig. lOA. The enclosure 15 part comprises a first control code detector 311, a second con- -trol code detector 312, flip-flops 321 and 322, gates 287, 331 and 332, a one-dimensional coder 234, and decoded signal memories 341 and 342..
The following will describe the construction and the oper-ation of the decoding device of Fig. 15 in detail. A codedsignal from the input terminal 201 i~ once stored in the input buffer memory 2G2. The signal from the input buffer memory 202 is checked first by the first and second control code detectors 311 and 312 as to whether the signal is the one-dimensional or two-dimensional coded one.
If the inputted control code is, for example "01111110", the signal is judged as the two-dimensional coded one, and the second control code detec~or 312 provides an output "1" to set the flip-flop 322, opening the gate 288. When the control code ; ' : ' , ~ , " ! "
', , ` ,; .,' ' ', ' ` ' , ,'' `': ' ~', ` , ~86~i is, for example, "01111111", the signal is judged as the one-dimensional coded signal, and the ~irs~ control cvde detector 311 yields an output "1" to set the flip-flop 321, opening the gate 287. At this time, the flip-flop 322 is reset ; consequ-ently, the gate 288 is cut off.
In a case of the two-dimen~ional coded signal being applied to open the gate 288, the mode code identify circuit 203, which has such a construction as shown in Fig. 5B, responds to open-ing of the gate 288 to read a required number of signals (for example, four bits at most, as shown in Table 1) from the input buffer memory 202, identifying the mode of the input signal, i.e.
any of the Pass mode, the Horizontal mode and the Vertical mode.
When the signal is "1110", it is regarded as indicating the Pass mode, and "1" is outputted on a line p ; when the signal is "1111", it is regarded as indicating the Horizontal mode, and "1" is provided on a line h ; when the signal is "0", "lQ0" or "1100", it is regarded as indi¢ating that the direction of the distance blai is plus in the Vertical mode, and "1" is produced on a line V+ ; and when the signal is "101" or "1101" it is regarded as indicating that the direction of the distance bla, is minus in the Vertical mode, and "1" is yielded on a line V-.
The address control circuit 221 has such a construction as de-picted in Fig. 5C, from which when any one of the outputs p, V- and V+ from the mode code identify circuit is "1", pulses provided from SaO are applied to the memory 211 to shift it bit by bit from the aO address.
When the identify circuit 203 provides "1" on the line p, the address control circuit 221 shifts the reference line memory 211 from the address of the picture element aQ to start detection 6~5i of the picture elements bl and b2. The reference line memory 211 has prestored therein information of the previous line via the decoded line memory 21~. The change picture element detector 240 has the construction shown in Fig. 4B and provides an output "1" upon each detection of a picture element differ-ent from the immediately preceding one in the signal train applied from the line memory 211. At th~ moment when the change picture element detector 240 p.rovides the output "1", if the detected picture element is differen~ in level from the picture element aO r the output "1" is applied via the exclusive OR cir-cuit 293 to the bl detector (an AND circuit) 251 to produce an output "1" on a line blp. The aOb, counter 272 receives pulses from the address control circuit 221 and counts the number of pulses occurring in the time interval f~om the aQ address to b lS lunit "1" is provided on the line blp). The b2 detector 252 outputs "1" on a line b2p when another change picture element is detected by the change picture element detector 240 after detection of the picture element b2 ("1" on the ~ine blp).
This b~ detector comprises a flip-flop and an AND circuit.
The a0b2 counter 271 receives pulses from the address control circuit 221 and counts them occurring in the time interval from the aO address to b2 (until "1" is provided on the line b2p).
Upon occurrence of "1" on the line b2p, the addres~ control cir-cuit 221 once stops sending out of the shift pulses. The in-forma~ion of the a0b2 counter 271 is applied to the aO register300 via the gate 281, which is opened by the provisiGn of the output "1" on the line p of the mode code identify circuit 203.
The contents of the aO register 300 are added to the address control circuits 221 and 222, so that the aO address is newly .. . . . . .

~2~64~S

set and the decoding operation is resumed.
In a case where the ldentify circuit 203 provides "l" on the line V+ or v~ (Vertical mode), the output "l" from the OR
circuit 291 is applied to the address control circuit 221 and the blal decoder 231. As a consequence, decoding relating to the abovesaid bl and b2 takes place, and the count value of the aObl counter indicates the address of the picture element bl relative to the picture element aO.
The b~al decoder 231 reads signals of one word from the input buffer memory 202 and decodes them. The decoded value is added by the adder 261 to the value of the aObl counter 272 and, at the same time, subtracted by the subtractor 262 from the value of the aOb~ coun~er 272. In a case where the output line V+
of the mode code identify circuit 203 is "l", the gate 284 is opened, so that ~he contents of the adder 261 is provided via the OR circuit 292 to the address control circuit 222 and to the aO register 300 via the gate 282. In contrast thereto, if the output line V- of the mode code identify circuit 203 is "l", the gate 285 is opened, passing the contents of the subtractor 262 to the address control circuit 222 vi~ the OR
circuit 292 and to the aO register 300 via the gate 282.
The address control circuit 222 has such a construction as depicted in Fig. lOB, which sets up the address of the picture element al on the basis of the contents transmitted thereto via the OR circuit 292, reproduces the picture element signals on the decoded line from the picture element aO to a picture element immediately preceding al identical with the level of the picture element aO and inverts the level of the picture element al relative to the level of the picture element - ~8 -:~ , .2~

aO. The content of the aO registe~ 300 is applied to the address control circuits 221 and 222, newly setting the address of the picture element aO and resuming decoding.
In a case where the line h of the mode code identify cir-cuit 203 becomes 1 (Horizontal mode), ~he aOal and a~a2 de-coders 232 and 233 successively read signals of two words from the input buffer memory 202 and the aOal decoder 232 decodes the first one word and the a~a2 decoder 233 the second one word. The decoded values are added to the address control cir-cuit 222 and to the aO register 300 via the gate 283 or 286.The address con~rol circuit 222 sets up ths addresses of the picture elements a, and aO, reproduces the picture element sig-nal on the decoded line from the picture element aO to a picture element immediately preceding al to be the same level as that of the picture element aO and inverts the level of the picture element a, and, thereafter, reproduces the picture element sig-nals from the picture element al to a picture element i~medi-ately preceding a, to be the same level as that of the picture element a, and sets the level of the picture element a2 to be different from the level of the picture element a,. The aO
address register 300 restores the addresses of the picture elements a, and a2, so that the a2 address becJlmes a new aO ad-dress. This new information is provided to the address control circuits 221 and 222 to set the aO address and restart decoding.
The two-dimensional decoded outputs of the Vertical and Horiæontal modes thus applied to the address control circuit 222 is processed therein as described above and then stored in the decoded signal memory 342. In this case, since the flip-flop 322 is in the set state, the gate 332 is opened by its - ~ 49 -1 " i :

~Z86~

output, so that the two-dimensional decoded signal stored in the decoded signal memory 342 is applied to the decoded line memory 212 and then outputted via the output terminal 350.
Next, when the ~irst control code detector 311 detects 5 the control code indicating the one-dimensional coded signal, the gate 287 is opened, as mentioned abov~, and the signal of the line is decoded by the one-dimensional decoder 234, there-after being stored in the decoded signal memory 341. At this time, since the gate 331 is open, the one-dimensional decoded 10 signal is provided to the decoded line memory 212, thereafter being outputted via the output terminal 350.
Also in respect of the above decoding device, the reset-ting conditions for the detectors, the registers, the counters and so forth have been neither described nor shown in the 15 drawings; but required ones of them (the mode code identify circuit 203, the b2 detector 252, the address control circuits 221 and 222, the counters 271 and 272, the decoders 231, 232 and 233, etc.) are reset for each setting of the aO address.
The texmination of one line is achieved by supervising the aO
20 address with the address control circuit 222 and, at the momen~
of the address of the picture element aO becoming the address of the last picture element of a scanning line, decoding of that line is completed and decoding of the next line is res~ned.
In the embodiment described above, the numbers of bits 25 of the one-dimensional and two-dimensional coded signals for each line are compared, and the coded signal of a smaller num-ber of coded bits is selected; but this comparison between the amounts of information of the one-dimensiollal and two-dimensional coded signals is not limited specifically to the ~1286~5 above. For example, the absolute number and a predetermined reference number of picture element changing points of the line to be coded are compared with each other ; if the former is ~maller than the latter, the one-dimensional coded line is used, and if the latter is smaller than the former, the two-dimensional coded line is used. Similarly, a difference between the absolute number of picture element changing points of the line to be coded and the absolute number of picture ele-ment changing points of an immediately preceding reference line i8 compared with a predetermined reference number ; if the former is smaller than the latter, the two-dimensional coded line is used, and if the former is larger than the latter, the one-dimensional coded line is used.
In the above, the one-dimensional and two-dimensional coded lines are selectively employed in accordance with the results of comparison between the amounts of information of the one-dimen~ional and two-dimensional coded signals at the end of scanning of one line, but it is also possible to perform coding and comparison for each signal of a predetermined length on one scanning line. Moreover, while the above embodiment has been d~scribed in connection with a case of using the two-dimensional sequential coding system, the invention can be carried into practice even if some other tw~-dimensional cod-ing system is used.
As described in the foregoing, according to this invention, a digital facsimile signal is coded by the one-dimensional and the two-dimensional coding system for each line and, in accord-ance with the amounts of information of the two coded signals, a more favorable one of them is selected as a coded ou~put, ~2~3S4S

for example, as shown in Fig. 16. Accordingly, there is the possibility that two-dimensional coded outputs are sucessively produced over a number of lines. With the two-dimensional coding system, however, ~ach line is coded and decoded utiliz-ing picture si~nal in~ormation of a refer~nce li~e immediatelypreceding it, as described previously, and a code error result-ing from a circuit noise or the like is likely to lead to a substantial degradation of the picture quality of reproduced pictures in those lines following that in which the code error has occurred. Therefore, in a case where when a code error is detected, a request repeat system can be used as in a four-wire private circuit or data communication network and a two-wire network circuit like an ordinary telephone circuit is employed, it is necessary to prevent spreading of the error.
Next, a description will be given of a system for limit-ing degradation of the picture quality of a reproduced picture due to the code error. This is to prevent that in the one~
dimensional, two-dimensional adaptive coding system described in the foregoing, the number of two-dimensional coded lines being outputted in succession sxceeds, for example, K lines (K is selected suitably but is shown to be five,), as shown in Fig. 16.
In Fig. 16, in a case where it is judged that ~ one-dimensional coded line i5 favorable for a first line and that two-dimansional coded lines are favorable for second to eighth lines, a one-dimensional coded line is cGmpulsorily used for the sixth line instead of the two-dimensional coded line so that K does not exceed five. In Fig. 16, for a ninth line, a one-dimensional coded output is produced according to the judgement ~ 52 ~

~8~5 that it is favorable for the lineO Even if the one-dimensional coded line is selected as a result of comparison between the one-dimensional and t~70-dimensional coded lines, a one-dimensional coded line is thus compulsoril~ inserted after K-l successive ~wo-dimensional coded lines counting from the one-dimensional coded line. Accordingly, a one-dimensional coded line may in some cases be inserted after two-dimensional lines less than K are output~ed.
In an embodiment of this invention based on such princi-ples, there are provided in the coding device a scale-of-K
counter 130, an inhibit circuit 131 and OR circuit 132, as indi-cated by the broken line in Fig. 17. When the output S2 from the comparator 62 is produced successively for K lines, the output S2 is inhibited by the inhibit circuit 131, and the out-put from the OR circuit 132 is applied to the first control code generator 102 and the gate 78, with the result that the first control code and a one-dimensional coded signal are transferred to the signal combiner 110. For the decoding device, howe~er, no modification is needed.
As has been described in detail in the foregoing, th~
present invention permits a substantial reduction of the amount of information to be transmitted and prevents spreading of degraded picture qualit~ due to a code error or the like.

.

Claims (8)

What we claim is :

1. A transmission method for a facsimile signal, in which a two-level facsimile signal obtained by scanning an original picture and successively sampling the scanning output into picture elements is received, as an input, and in which the position of an information change picture element having changed from one to the other of two signal levels is coded and sent out, the improvement of the method comprising :
a first step of setting a starting picture element on a coding scanning line to be coded from which the coding starts ;
a second step of detecting a first information change picture element lying next to the starting picture element on the coding scanning line ;
a third step of detecting a first reference picture ele-ment, which is a first information change picture element lying after a picture element just above the starting picture element on a reference scanning line immediately preceding the coding scanning line and has a signal level different from that of the starting picture element, and a second reference picture element of an information change picture element next to the first reference picture element ;
a fourth step of detecting, as a first mode, the state in which the second reference picture element precedes a pic-ture element just above the first information change picture element by more than n (n being O or a positive integer) picture elements ;
a fifth step of detecting, as not the first mode, the state in which the second reference picture element does not precede a picture element just above the first information change picture element by more than n picture elements ;
a sixth step of comparing a first correlation between the starting picture element and the first information change picture element with a second correlation between the first information change picture element and the first reference picture element when the abovesaid state is detected as not the first mode ;
a seventh step of coding the presence of the first and second reference picture elements as the first mode and setting the picture element just below the second reference picture element as the starting picture element in the first step when the first mode is detected ;
an eighth step of coding a distance between the starting picture element and the first information change picture element-as a second mode and setting the first information change picture element as the starting picture element in the first step when the first correlation is higher than the second correlation ;
a ninth step of coding the distance between the first in-formation change picture element and the first reference picture element as a third mode and setting the first information change picture element as the starting picture element in the first step when the first correlation is not higher than the second correlation ; and a tenth step of sending out the coded outputs of the seventh, eighth and ninth steps after combining them into a com-posite signal of two-dimensional codes.

2. A transmission method for a facsimile signal according to claim 1, further including ;
an eleventh step of successively coding by a one-dimensional method information change picture elements on a coding scanning line to be coded for each predetermined length of the coding scanning line to develop one-dimensional codes and storing the one-dimensional codes ;
a twelfth step of comparing the information amount of the one-dimensional codes and the two-dimensional codes stored for each predetermined length of the coding scanning line ;
a thirteenth step of selecting said composite signal as an output when the information amount of the one-dimensional codes is higher than the information amount of the two-dimensional codes ;
a fourteenth step of selecting the one-dimensional codes as an output when the information amount of the one-dimensional codes is not higher than the information amount of the two-dimensional codes ; and a fifteenth step of adding a peculiar control code to the coded output of each of the thirteenth and fourteenth steps for sending out them after combining into a composite trans-mission signal.

3. A transmission method for a facsimile signal, in which a two-level facsimile signal obtained by scanning an original picture and successively sampling the scanning output into pic-ture elements is received as an input, and in which the posi-tion of an information change picture element having changed from one to the other of two signal levels is coded and sent out, the improvement of the method comprising :
a first step of setting a starting picture element on a coding scanning line to be coded from which the coding starts;
a second step of detecting a first information change pic-ture element lying next to the starting picture element on the coding scanning line ;
a third step of detecting a first reference picture ele-ment, which is a first information change picture element lying after a picture element just above the starting picture element on a reference scanning line immediately preceding the coding scanning line and has a signal value different from that of the starting picture element, and a second refer-ence picture element which is an information change picture element next to the first information change picture element;
a fourth step of detecting, as a first mode, the state in which the second reference picture element precedes a pic-ture element just above the first information change picture element by more than n (n being 0 or a positive integer) pic-ture elements ;
a fifth step of detecting, as not the first mode, the state in which the second reference picture element does not precede a picture element just above the first information change picture element by more than n picture element ;
a sixth step of comparing a first correlation between the starting picture element and the first information change picture element with a second correlation between the first information change picture element and the first reference picture element when the abovesaid state is detected as not the first mode ;
a seventh step of coding the presence of the first and second reference picture elements as the first mode and set-ting the picture element just below the second reference pic-ture element as the starting picture element in the first step when the first mode is detected ;

an eighth step of coding a distance between the start ing picture element and the first information change picture element as a second mode and setting the first information change picture element as the starting picture element in the first step when the first correlation is higher than the second correlation ;
a ninth step of coding a distance between the first in-formation change picture element and the first reference pic-ture element as a third mode and setting the first information chnage picture element as the starting picture element in the first step when the first correlation is not higher than the second correlation ;
a tenth step of temporarily stopping the two-dimensional coding operation evey time the number of coding scanning lines has reached a predetermined value and coding the positions of information change picture elements of the next coding scan-ning line only without referring to the positions of infor-mation change picture elements of another scanning line ; and an eleventh step of sending out the coded outputs of the seventh, eighth, ninth and tenth steps after combining them into a composite signal of two-dimensional codes.

4. A transmission method for a facsimile signal according to claim 3, further including :
a twelfth step of successively coding by a one-dimensional method information change picture elements on a coding scanning line to be coded for each predetermined length of the coding scanning line to develop one-dimensional codes and storing the one-dimensional codes ;
a thirteenth step of comparing the information amount of the one-dimensional codes and the two-dimensional codes stored for each predetermined length of the coding scanning line ;
a fourteenth step of selecting said composite signal as an output when the information amount of the one-dimensional codes is higher than the information amount of the two-dimen-sional codes ;
a fifteenth step of selecting the one-dimensional codes as an output when the information of the one-dLmensional codes is not higher than the information amount of the two-dimensional codes ; and a sixteen step of adding a peculiar control code to the coded output of each of the fourteenth and fifteenth steps for sending out them after combining into a composite transmission signal.

5. A transmission method for a facsimile signal in which a two-level facsimile signal obtained by scanning an original picture and successively sampling the scanning output into picture elements is received as an input, and in which the position of an information change picture element having changed from one to the other of two signal levels is coded and sent out, the improvement of the method comprising :
a first step of setting a starting picture element on a coding scanning line to be coded from which the coding starts ;
a second step of detecting first and second information change picture elements sequentially lying after the starting picture element on the coding scanning line ;
a third step of detecting a first reference picture ele-ment, which is a first information change picture element lying after a picture element just above the starting pictuxe element _ 59 _ .

on a reference scanning line immediately preceding the cod-ing scanning line and has a signal value different from that of the starting picture element, and a second reference pic-ture element which is an information change picture element next to the first information change picture element ;
a fourth step of detecting, as a first mode, the state in which the second reference picture element precedes a pic-ture element just above the first information chnage picture element by more than n (n being 0 or a positive integer) pic-ture elements ;
a fifth step of detecting, as not the first mode, the state in which the second reference picture element does not precede a picture element just above the first information change picture element by more than n picture elements ;
a sixth step of comparing a first correlation between the starting picture element and the first information change pic-ture element with a second correlation between the first in-formation change picture element and the first reference pic-ture element when the abovesaid state is detected as not the first mode ;
a seventh step of coding the presence of the first and second reference picture elements as the first mode and setting the picture element just below the second reference picture element as the starting picture element in the first step when the first mode is detected ;
an eighth step of coding a distance between the starting picture element and the first information change picture ele-ment and a distance between the first and second information change picture elements as a second mode and setting the second information change picture element as the starting picture element in the first step when the first correlation is higher than the second correlation ;
a ninth step of coding a distance between the first in-formation change picture element and the first reference pic-ture element as a third mode and setting the first information change picture element as the starting picture element in the first step when the first correlation is not higher than the second correlation ; and a tenth step of sending out the coded outputs of the seventh, eighth and ninth steps after combining them into a composite signal of two-dimensional codes.

6. A transmission method for a facsimile signal according to claim 5, further including :
an eleventh step of successively coding by a one-dimen-tional method information change picture elements on a coding scanning line to be coded for each predetermined length of the coding scanning line to develop one-domensional codes and storing the one-dimensional codes ;
a twelfth step of comparing the information amount of the one-dimensional codes and the two-dimensional codes stored for each predetermined length of the coding scanning line ;
a thirteenth step of selecting said composite signal as an output when the information amount of the one-dimensional codes is higher than the information amount of the two-dimen-tional codes ;
a fourteenth step of selecting the one-dimensional codes as an output when the information of the one-dimensional codes is not higher than the information amount of the two-dimensional codes ; and a fifteenth step of adding a peculiar control code to the coded output of each of the tenth, thirteenth and four-teenth steps for sending out them after combining into a composite transmission signal.

7. A transmission method for a facsimile signal, in which a two-level facsimile signal obtained by scanning an original picture and successively sampling the scanning output into picture elements is received as an input, and in which the position of an information change picture element having changed from one to the other of two signal levels is coded and sent out, the improvement of the method comprising :
a first step of setting a starting picture element on a coding scanning line to be coded from which the coding starts;
a second step of detecting first and second information change picture elements sequentially lying after the starting picture element on the coding scanning line ;
a third step of detecting a first reference picture ele-ment, which is a first information change picture element lying after a picture element just above the starting picture element on a reference scanning line immediately preceding the coding scanning line and has a signal level different from that of the starting picture element, and a second refer-ence picture element which is an information change picture element next to the first reference change picture element ;
a fourth step of detecting, as a first mode, the state in which the second reference picture element precedes a pic-ture element just above the first information change picture element by more than n (n being O or a positive integer) picture elements ;
a fifth step of detecting, as not the first mode, the state in which the second reference picture element does not precede a picture element just above the first information change picture element by more than n picture elements ;
a sixth step of comparing a first correlation between the starting picture element and the first information change picture element with a second correlation between the first information change picture element and the first reference picture element when the abovesaid state is detected as not the first mode ;
a seventh step of coding the presence of the first and second reference picture elements as the first mode and setting the picture element just below the second reference picture element as the starting picture element in the first step when the first mode is detected ;
an eighth step of coding a distance between the starting picture element and the first information change picture ele-ment and a distance between the first and second information change picture elements as a second mode and setting the second information change picture element as the starting picture element in the first step when the first correlation is higher than the second correlation ;
a ninth step of coding a distance between the first in-formation change picture element and the first reference pic-ture element as a third mode and setting the first information change picture element as the starting picture element in the first step when the first correlation is lower than the second correlation ;

a tenth step of temporarily stopping the two-dimensional coding operation every time the number of coding scanning lines has reached a predetermined value and coding the positions of information change picture elements of the next coding scan-ning line only without referring to the positions of informat-ion change picture elements of another scanning line ;
an eleventh step of sending out the coded outputs of the seventh, eighth, ninth and tenth steps after combining them into a composite signal of two-dimensional codes.

8. A transmission method for a facsimile signal according to claim 7, further including :
a twelveth step of successively coding by a one-dimensional method information change picture elements on a coding scanning line to be coded for each predetermined length of the coding scanning line to develop one-dimensional codes and storing the one-dimensional codes ;
a thirteenth step of comparing the information amount of the one-dimensional codes and the two-dimensional codes stored for each predetermined length of the coding scanning line ;
a fourteenth step of selecting said composite signal as an output when the information amount of the one-dimensional codes is higher than the information amount of the two-dimen-tional codes ;
a fifteenth step of selecting the one-dimensional codes as an output when the information of the one-dimensional codes is not higher than the information amount of the two-dimensional codes; and a sixteenth step of adding a peculiar control code to the coded output of each of the eleventh, fourteenth and fifteenth steps for sending out them after combining into a composite transmission signal.