CA2178961C - Method, post-processing filter, and video compression system for suppressing mosquito and blocking artifacts - Google Patents
Method, post-processing filter, and video compression system for suppressing mosquito and blocking artifacts Download PDFInfo
- Publication number
- CA2178961C CA2178961C CA002178961A CA2178961A CA2178961C CA 2178961 C CA2178961 C CA 2178961C CA 002178961 A CA002178961 A CA 002178961A CA 2178961 A CA2178961 A CA 2178961A CA 2178961 C CA2178961 C CA 2178961C
- Authority
- CA
- Canada
- Prior art keywords
- post
- intensity
- predetermined
- pixel
- pixels
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/527—Global motion vector estimation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
- H04N19/86—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving reduction of coding artifacts, e.g. of blockiness
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
Abstract
The visibility of mosquito and blocking artifacts in a video compression system are effectively suppressed by calculating a postprocessed pixel intensity (300). A post-processed pixel intensity is determined using a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a local neighbourhood of pixels (308) that includes a plurality of previously post-processed pixel intensities and a plurality of original pixel intensities. This eliminates the dependency on local signal and noise power estimates.
Description
2 1 78~6 1 METHOD, POST-PROCESSING FILTER, AND VIDEO
COMPRESSION SYSTEM FOR SUPPRESSING MOSQUITO AND
BLOCKING ARTIFACTS
Field of the Invention The present invention relates generally to video compression systems, and more particularly to suppression of 10 mosquito and blocking artifacts in a video compression system.
Background In video compression systems, the number of bits required to represent a single frame in a video sequence can be reduced by taking advantage of the similarities that the single frame has with previous frames. Methods such as a hybrid block motion compensated discrete cosine transform, DCT, encoding scheme have been used to achieve very high compression ratios, but these methods also introduce artifacts which severely degrade the visual quality of the decoded sequence.
The two types of artifacts introduced are mosquito artifacts and blocking artifacts. Mosquito artifacts are defined as temporally nonstationary impulses that appear around objects which are moving within the decompressed video. These artifacts result from the coarse quantization of the prediction error signal. The majority of the energy contained in a prediction error signal is the result of the motion estimator's inability to distinguish between differently moving objects. For instance, in video conferencing applications the subject is generally against a 35 stationary background. Since the motion estimator tries to WO 96/14708 ~ gS/11187 21 7~61 match blocks of pixels between temporally adjacent frames, the boundaries between moving objects and stationary background that fall within these blocks cannot be detected.
This leads to the situation where either part of the background 5 is assumed to be moving, or part of the moving object is assumed to be stationary. Coarsely quantizing these prediction errors results in impulsive artifacts that change over time and tend to swarm around the moving object, similar in manner to a mosquito.
Blocking artifacts are defined as the introduction of artificial block boundaries into the decoded video sequence.
These artifacts are due to the combination of dividing the prediction error signal into blocks, as well as quantization.
15 That is, since there exists an inverse relationship between spatial extent and frequency extent analogous to the inverse relationship that exists between time and frequency extent in Fourier analysis, the quantization errors that occur in the DCT
domain are smeared across the corresponding spatial block.
20 Furthermore, since each block is quantized separately, the errors are most visible at the block boundaries.
Since both of these artifacts occur at specific locations and not uniformly throughout the video sequence, post-2 5 processing filters, designed to reduce both artifacts, must bespatially adaptive. Spatially adaptive filters have been used to suppress these artifacts, but they rely on local signal and noise power estimates to alter their responses. Several problems exist with this design. For instance, estimation of 30 the noise power based on the quantization step size is not a reliable indicator as to the spatial location of mosquito and blocking artifacts within the decompressed video. Over-smoothing or blurring of the decompressed video occurs due to inaccurate estimates of the decompressed video's signal-to-35 noise ratio, SNR. While enhancement of the blocking artifacts WO96/14708 2 1 7 8 9 6 1 PCr/USgS/11187 can result from the incorrect classification of blockboundaries as object edges within the decompressed video.
The effect of any one of these problems on the perceived visual quality of the decompressed video is disastrous.
A further consideration is that users of compressed video demand that the decoder be able to resize the video display window to any specified size. To provide a user with this capability, requires the decoder to perform an additional, 10 with regards to removing coding artifacts, post-processing operation involving data interpolation or decimation. An important requirement is that this resize operation must be computationally efficient and provide a resize sequence that is visually appealing. That is, artifacts should not be introduced 15 into the video sequence as a result of the resizing operation.
- Furthermore, coding artifacts should not become more visible in the decoded sequence due to the resizing operation.
Thus, there is a need for a method, post-processing 20 filter, and video compression system which suppress mosquito and blocking artifacts independent of local signal to noise power ratio estimates, while also providing the ability to resize the decompressed video.
Brief Description of the Drawings FIG. 1 is a flow chart of steps implemented by a post-processing filter which suppresses mosquito and blocking 3 0 artifacts in accordance with the present invention.
FIG. 2 is a block diagram of a video compression system that uses a post-processing filter which suppresses mosquito and blocking artifacts in accordance with the present 3 5 invention.
WO 96/14708 PCT/USg5/11187 FIG. 3 is a block diagram of a post-processing filter which suppresses mosquito and blocking artifacts in accordance with the present invention.
FIG. 4 is a block diagram of a post-processing filter which suppresses mosquito and blocking artifacts, using a predictor and an updater, in accordance with the present Invention.
FIG. 5 is a diagram of a post-processing filter which suppresses mosquito and blocking artifacts and performs resizing in accordance with the present invention.
Detailed Description of a Preferred Embodiment The present invention allows a post-processing filter in a video compression system to suppress mosquito and blocking 20 artifacts and to be independent of local signai and noise power estimates. A signal variance is estimated based on a local neighborhood of pixels. The local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of original pixel intensities. An 25 optimal filter gain is calculated based on the signal variance, a predetermined noise variance, and a predetermined autocorrelation coefficient. A post-processed pixel intensity is determined using the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal 30 filter gain. Also, if desired, a post-processed resized pixel intensity is determined using the predetermined autocorrelation coefficient, the local neighborhood of pixels, the optimal filter gain, and plurality of interpolation coefficients. This eliminates the dependency on local signal to WOg6/14708 ;~1 7~96~ P~ 3S/11187 .
noise power ratio estimates, whiie providing the user with the ability to resize and post-process the decompressed video.
The present invention is more fully described with 5 reference to FlGs. 1 - 5. FIG. 1, numeral 100, is a flow chart of steps implemented by a post-processing filter which suppresses mosquito and blocking artifacts in accordance with the present invention. First, a local neighborhood of pixels is created, in a memory unit, based on a decompressed video sequence (101). Then, a signal variance is estimated based on a local neighborhood of pixels (102). An optimal filter gain is then calculated using the signal variance, a predetermined noise variance, and a predetermined autocorrelation coefficient (104). Finally, a post-processed pixel intensity is determined based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain (106).
When resizing of the decompressed video window is 2 0 required, the post-processed pixel intensity is a resized post-processed pixel intensity based on a plurality of predetermined interpolation coefficients. This is accomplished by combining the resizing operation with the post-processing filter described above. The post-processed pixels are removed from the local neighborhood and replaced with decompressed pixels.
The parameters described above, the signal variance based on a local neighborhood of pixels and the optimal filter gain based on the signal variance, a predetermined noise variance, and a predetermined autocorrelation coefficient, are combined with 3 a plurality of interpolation coefficients to form the coefficients of this combined resize post-processed filter.
Finally, a resized post-processed pixel intensity is determined based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, the optimal filter gain, and 3 5 plurality of interpolation coefficients.
W O 96/14708 PC~rnUS9Stlll87 21/~S~1 6 The signal variance may be estimated by using an equation of the form:
cr2(n) = 1 [ ~(x(n _ i))2 + M (y(n - i))2] ~ m 2(n) where mS(n) = 2U I [ ~; x(n - i) + ~y(n - i)]
.
.
10 is the average intensity value in the (2 M + l) local neighborhood of pixels, y(n) iS a previously post-processed pixel intensity in the local neighborhood of pixels, and x(n) is a decompressed pixel intensity in the local neighborhood of pixels. The optimal filter gain may be calculated by using an equation of the form:
~ + a ~S2 (n) where - I 2 ~
+ ~2 2 0 and where ~n2 is the noise variance, cr~(n) is the signal variance at pixel location n, and ~ is the predetermined autocorrelation coefficient. A typical range for the noise variance is 0~<10. A typical range for the autocorrelation coefficient is 0.9~<1.0,. The post-processed pixel intensity may be 2 5 based on an estimate of the intensity of a pixel. The estimate of the intensity of a pixel is a function of the predetermined autocorrelation coefficient, a decompressed pixel intensity of a next pixel relative to a predetermined scanning direction, and an immediately previously post-processed pixel intensity.
WO 96/14708 2 1 7 ~ 9 6 ~ /11187 The estimate of the intensity of a pixel may be determined by using an equation of the form:
jO(n)=~-[9b(n-l)+~(n+l)]
where yO(n) is the estimate of the intensity of a pixel, ~ is the predetermined autocorrelation coefficient, x(n+l) is the decompressed pixel intensity of a next pixel relative to a predetermined scanning direction, and yb(n-l) is an 10 immediately previously post-processed pixel intensity. The post-processed pixel intensity may determined by using an equation of the form: -yb(n)=ya(n)+G(n)-[x(n)-yo(nj]
where C(n) is the optimal filter gain, yO(n) is the estimate of the intensity of a pixel in the local neighborhood of pixels, and x(n) is the decompressed pixel intensity in the local neighborhood of pixels.
For the generalized resizing filter of the form:
y(m)= Co x(n)+ Cl x(n +I)+ C ~ x(n -1)+ C2 x(n + 2)+ C 2 x(n - 2)+.. .
2 5 where y(m) iS the resized decompressed pixel located between x(n) and x(n +l) and C~ is a predetermined interpolation coefficient which is dependent on the interpolation or decimation method, the following equation is used to determine the resized post-processed pixel:
WO 96/14708 1 ~ b~llllM
yb(m) = [c-l g(n)+ Co (l - g(n))-B]x(n -l)+[co-g(n)+(l- g(n)) (C_~ ~ + C~-B~x(n) +[C~-g(n)+(l - g(n)) (C2 ~ + CO-B]x(n +l) + [C2 g(n)+ Cl (1- g(n))-Blx(n + 2) +C-2 x(n - 2)+ C3 x(n + 3)+ C-3 x(n - 3)+-..-where B=~
5 and yb(m) is the resized post-processed pixel located between x(n) and x(n + 1).
FIG. 2., numeral 200, is a block diagram of a video compression system that uses a post-processing filter which 10 suppresses mosquito and blocking artifacts in accordance with the present invention. The video compression system (200) includes a video encoding unit (204), a channel encoding unit (208), a channel (210), a channel decoding unit (212), a video decoding unit (214), and a post-processing unit (218) which 1 5 includes a post-processing filter (222).
Redundant information contained within the input video sequence (202) is removed by the video encoding unit (204) generating the compressed video stream (206). The 20 compressed video stream (206) is then encoded by the channel encoding unit (208) for efficient transmission over the channel (210) to the corresponding channel decoding unit (212). The channel decoding unit (212) decodes the received information and provides the video decoding unit (214) with the 25 compressed video stream (213). Based on the compressed video stream (213), the video decoding unit (214) generates the decompressed video sequence (216). This decompressed video sequence (216) is the input to the post-processing unit (218), where the post-processing filter (222) is employed to 30 remove the mosquito and blocking artifacts from the wog6/l4708 2 1 7 8 9 6 1 ~ 3S/11187 decompressed video resulting in the post-processed video sequence (220).
FIG. 3, numeral 300, is a block diagram of a post-5 processing filter which suppresses mosquito and blocking artifacts in accordance with the present invention. The post-processing filter (222) comprises a memory unit (302), a signal variance determiner (310), an optimal filter gain determiner (314), and a combining circuit (318). The memory 10 unit (302) receives and stores a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a decompressed video sequence (216). The decompressed video sequence (216) is stored as a local neighborhood of pixels (308).
The predetermined noise variance stored in the memory unit (302) is typically a value between 1 and 10. The signal variance determiner (310) estimates the signal variance (312) based on the local neighborhood of pixels (308). The optimal 2 0 filter gain determiner (314) determines an optimal filter gain (316) based on the signal variance (312), the predetermined noise variance (304), and the predetermined autocorrelation - coefficient (306). The combining circuit (318) determines the post-processed pixel intensity (320) based on the 2 5 predetermined autocorrelation coefficient (306), the local neighborhood of pixels (308), and the optimal filter gain (316).
The combining circuit (318) may be implemented as an infinite impulse response (IIR) filter or a finite impulse response (FIR) filter.
FIG. 4, numeral 400, is a block ~tagram of a post-processing filter which suppresses mos~uito and blocking artifacts, using a predictor (402) and an updater (404), in accordance with the present invention. The post-processing 35 filter comprises a memory unit (302), a signal variance 21 7~')61 1 o determiner (310), an optimal filter gain determiner (314), and a combining circuit (318). The memory unit (302) receives and stores a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a decompressed video 5 sequence (216). The decompressed video sequence (216) is stored as a local neighborhood of pixels (308). It should be noted that FIG. 4 reflects the case when M=2 which corresponds to the local neighborhood of pixels consisting of the five pixel elements indicated by the X's and O's. The choice 10 of M iS made at the time of implementation and is dependent on the amount of smoothness desired: the larger the M the more smoothing applied to the decompressed video. The X's - and O's indicate which pixels are previously post-processed intensity pixels, X's, and which are decompressed intensity 15 pixels, O's, assuming a left to right movement across each row of the image frame. Other methods can also be used for moving through the decompressed image frame stored in memory, such as from right to left across each row, top to bottom or bottom to top across each column.
The general operation of the block diagram shown in FIG.
COMPRESSION SYSTEM FOR SUPPRESSING MOSQUITO AND
BLOCKING ARTIFACTS
Field of the Invention The present invention relates generally to video compression systems, and more particularly to suppression of 10 mosquito and blocking artifacts in a video compression system.
Background In video compression systems, the number of bits required to represent a single frame in a video sequence can be reduced by taking advantage of the similarities that the single frame has with previous frames. Methods such as a hybrid block motion compensated discrete cosine transform, DCT, encoding scheme have been used to achieve very high compression ratios, but these methods also introduce artifacts which severely degrade the visual quality of the decoded sequence.
The two types of artifacts introduced are mosquito artifacts and blocking artifacts. Mosquito artifacts are defined as temporally nonstationary impulses that appear around objects which are moving within the decompressed video. These artifacts result from the coarse quantization of the prediction error signal. The majority of the energy contained in a prediction error signal is the result of the motion estimator's inability to distinguish between differently moving objects. For instance, in video conferencing applications the subject is generally against a 35 stationary background. Since the motion estimator tries to WO 96/14708 ~ gS/11187 21 7~61 match blocks of pixels between temporally adjacent frames, the boundaries between moving objects and stationary background that fall within these blocks cannot be detected.
This leads to the situation where either part of the background 5 is assumed to be moving, or part of the moving object is assumed to be stationary. Coarsely quantizing these prediction errors results in impulsive artifacts that change over time and tend to swarm around the moving object, similar in manner to a mosquito.
Blocking artifacts are defined as the introduction of artificial block boundaries into the decoded video sequence.
These artifacts are due to the combination of dividing the prediction error signal into blocks, as well as quantization.
15 That is, since there exists an inverse relationship between spatial extent and frequency extent analogous to the inverse relationship that exists between time and frequency extent in Fourier analysis, the quantization errors that occur in the DCT
domain are smeared across the corresponding spatial block.
20 Furthermore, since each block is quantized separately, the errors are most visible at the block boundaries.
Since both of these artifacts occur at specific locations and not uniformly throughout the video sequence, post-2 5 processing filters, designed to reduce both artifacts, must bespatially adaptive. Spatially adaptive filters have been used to suppress these artifacts, but they rely on local signal and noise power estimates to alter their responses. Several problems exist with this design. For instance, estimation of 30 the noise power based on the quantization step size is not a reliable indicator as to the spatial location of mosquito and blocking artifacts within the decompressed video. Over-smoothing or blurring of the decompressed video occurs due to inaccurate estimates of the decompressed video's signal-to-35 noise ratio, SNR. While enhancement of the blocking artifacts WO96/14708 2 1 7 8 9 6 1 PCr/USgS/11187 can result from the incorrect classification of blockboundaries as object edges within the decompressed video.
The effect of any one of these problems on the perceived visual quality of the decompressed video is disastrous.
A further consideration is that users of compressed video demand that the decoder be able to resize the video display window to any specified size. To provide a user with this capability, requires the decoder to perform an additional, 10 with regards to removing coding artifacts, post-processing operation involving data interpolation or decimation. An important requirement is that this resize operation must be computationally efficient and provide a resize sequence that is visually appealing. That is, artifacts should not be introduced 15 into the video sequence as a result of the resizing operation.
- Furthermore, coding artifacts should not become more visible in the decoded sequence due to the resizing operation.
Thus, there is a need for a method, post-processing 20 filter, and video compression system which suppress mosquito and blocking artifacts independent of local signal to noise power ratio estimates, while also providing the ability to resize the decompressed video.
Brief Description of the Drawings FIG. 1 is a flow chart of steps implemented by a post-processing filter which suppresses mosquito and blocking 3 0 artifacts in accordance with the present invention.
FIG. 2 is a block diagram of a video compression system that uses a post-processing filter which suppresses mosquito and blocking artifacts in accordance with the present 3 5 invention.
WO 96/14708 PCT/USg5/11187 FIG. 3 is a block diagram of a post-processing filter which suppresses mosquito and blocking artifacts in accordance with the present invention.
FIG. 4 is a block diagram of a post-processing filter which suppresses mosquito and blocking artifacts, using a predictor and an updater, in accordance with the present Invention.
FIG. 5 is a diagram of a post-processing filter which suppresses mosquito and blocking artifacts and performs resizing in accordance with the present invention.
Detailed Description of a Preferred Embodiment The present invention allows a post-processing filter in a video compression system to suppress mosquito and blocking 20 artifacts and to be independent of local signai and noise power estimates. A signal variance is estimated based on a local neighborhood of pixels. The local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of original pixel intensities. An 25 optimal filter gain is calculated based on the signal variance, a predetermined noise variance, and a predetermined autocorrelation coefficient. A post-processed pixel intensity is determined using the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal 30 filter gain. Also, if desired, a post-processed resized pixel intensity is determined using the predetermined autocorrelation coefficient, the local neighborhood of pixels, the optimal filter gain, and plurality of interpolation coefficients. This eliminates the dependency on local signal to WOg6/14708 ;~1 7~96~ P~ 3S/11187 .
noise power ratio estimates, whiie providing the user with the ability to resize and post-process the decompressed video.
The present invention is more fully described with 5 reference to FlGs. 1 - 5. FIG. 1, numeral 100, is a flow chart of steps implemented by a post-processing filter which suppresses mosquito and blocking artifacts in accordance with the present invention. First, a local neighborhood of pixels is created, in a memory unit, based on a decompressed video sequence (101). Then, a signal variance is estimated based on a local neighborhood of pixels (102). An optimal filter gain is then calculated using the signal variance, a predetermined noise variance, and a predetermined autocorrelation coefficient (104). Finally, a post-processed pixel intensity is determined based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain (106).
When resizing of the decompressed video window is 2 0 required, the post-processed pixel intensity is a resized post-processed pixel intensity based on a plurality of predetermined interpolation coefficients. This is accomplished by combining the resizing operation with the post-processing filter described above. The post-processed pixels are removed from the local neighborhood and replaced with decompressed pixels.
The parameters described above, the signal variance based on a local neighborhood of pixels and the optimal filter gain based on the signal variance, a predetermined noise variance, and a predetermined autocorrelation coefficient, are combined with 3 a plurality of interpolation coefficients to form the coefficients of this combined resize post-processed filter.
Finally, a resized post-processed pixel intensity is determined based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, the optimal filter gain, and 3 5 plurality of interpolation coefficients.
W O 96/14708 PC~rnUS9Stlll87 21/~S~1 6 The signal variance may be estimated by using an equation of the form:
cr2(n) = 1 [ ~(x(n _ i))2 + M (y(n - i))2] ~ m 2(n) where mS(n) = 2U I [ ~; x(n - i) + ~y(n - i)]
.
.
10 is the average intensity value in the (2 M + l) local neighborhood of pixels, y(n) iS a previously post-processed pixel intensity in the local neighborhood of pixels, and x(n) is a decompressed pixel intensity in the local neighborhood of pixels. The optimal filter gain may be calculated by using an equation of the form:
~ + a ~S2 (n) where - I 2 ~
+ ~2 2 0 and where ~n2 is the noise variance, cr~(n) is the signal variance at pixel location n, and ~ is the predetermined autocorrelation coefficient. A typical range for the noise variance is 0~<10. A typical range for the autocorrelation coefficient is 0.9~<1.0,. The post-processed pixel intensity may be 2 5 based on an estimate of the intensity of a pixel. The estimate of the intensity of a pixel is a function of the predetermined autocorrelation coefficient, a decompressed pixel intensity of a next pixel relative to a predetermined scanning direction, and an immediately previously post-processed pixel intensity.
WO 96/14708 2 1 7 ~ 9 6 ~ /11187 The estimate of the intensity of a pixel may be determined by using an equation of the form:
jO(n)=~-[9b(n-l)+~(n+l)]
where yO(n) is the estimate of the intensity of a pixel, ~ is the predetermined autocorrelation coefficient, x(n+l) is the decompressed pixel intensity of a next pixel relative to a predetermined scanning direction, and yb(n-l) is an 10 immediately previously post-processed pixel intensity. The post-processed pixel intensity may determined by using an equation of the form: -yb(n)=ya(n)+G(n)-[x(n)-yo(nj]
where C(n) is the optimal filter gain, yO(n) is the estimate of the intensity of a pixel in the local neighborhood of pixels, and x(n) is the decompressed pixel intensity in the local neighborhood of pixels.
For the generalized resizing filter of the form:
y(m)= Co x(n)+ Cl x(n +I)+ C ~ x(n -1)+ C2 x(n + 2)+ C 2 x(n - 2)+.. .
2 5 where y(m) iS the resized decompressed pixel located between x(n) and x(n +l) and C~ is a predetermined interpolation coefficient which is dependent on the interpolation or decimation method, the following equation is used to determine the resized post-processed pixel:
WO 96/14708 1 ~ b~llllM
yb(m) = [c-l g(n)+ Co (l - g(n))-B]x(n -l)+[co-g(n)+(l- g(n)) (C_~ ~ + C~-B~x(n) +[C~-g(n)+(l - g(n)) (C2 ~ + CO-B]x(n +l) + [C2 g(n)+ Cl (1- g(n))-Blx(n + 2) +C-2 x(n - 2)+ C3 x(n + 3)+ C-3 x(n - 3)+-..-where B=~
5 and yb(m) is the resized post-processed pixel located between x(n) and x(n + 1).
FIG. 2., numeral 200, is a block diagram of a video compression system that uses a post-processing filter which 10 suppresses mosquito and blocking artifacts in accordance with the present invention. The video compression system (200) includes a video encoding unit (204), a channel encoding unit (208), a channel (210), a channel decoding unit (212), a video decoding unit (214), and a post-processing unit (218) which 1 5 includes a post-processing filter (222).
Redundant information contained within the input video sequence (202) is removed by the video encoding unit (204) generating the compressed video stream (206). The 20 compressed video stream (206) is then encoded by the channel encoding unit (208) for efficient transmission over the channel (210) to the corresponding channel decoding unit (212). The channel decoding unit (212) decodes the received information and provides the video decoding unit (214) with the 25 compressed video stream (213). Based on the compressed video stream (213), the video decoding unit (214) generates the decompressed video sequence (216). This decompressed video sequence (216) is the input to the post-processing unit (218), where the post-processing filter (222) is employed to 30 remove the mosquito and blocking artifacts from the wog6/l4708 2 1 7 8 9 6 1 ~ 3S/11187 decompressed video resulting in the post-processed video sequence (220).
FIG. 3, numeral 300, is a block diagram of a post-5 processing filter which suppresses mosquito and blocking artifacts in accordance with the present invention. The post-processing filter (222) comprises a memory unit (302), a signal variance determiner (310), an optimal filter gain determiner (314), and a combining circuit (318). The memory 10 unit (302) receives and stores a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a decompressed video sequence (216). The decompressed video sequence (216) is stored as a local neighborhood of pixels (308).
The predetermined noise variance stored in the memory unit (302) is typically a value between 1 and 10. The signal variance determiner (310) estimates the signal variance (312) based on the local neighborhood of pixels (308). The optimal 2 0 filter gain determiner (314) determines an optimal filter gain (316) based on the signal variance (312), the predetermined noise variance (304), and the predetermined autocorrelation - coefficient (306). The combining circuit (318) determines the post-processed pixel intensity (320) based on the 2 5 predetermined autocorrelation coefficient (306), the local neighborhood of pixels (308), and the optimal filter gain (316).
The combining circuit (318) may be implemented as an infinite impulse response (IIR) filter or a finite impulse response (FIR) filter.
FIG. 4, numeral 400, is a block ~tagram of a post-processing filter which suppresses mos~uito and blocking artifacts, using a predictor (402) and an updater (404), in accordance with the present invention. The post-processing 35 filter comprises a memory unit (302), a signal variance 21 7~')61 1 o determiner (310), an optimal filter gain determiner (314), and a combining circuit (318). The memory unit (302) receives and stores a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a decompressed video 5 sequence (216). The decompressed video sequence (216) is stored as a local neighborhood of pixels (308). It should be noted that FIG. 4 reflects the case when M=2 which corresponds to the local neighborhood of pixels consisting of the five pixel elements indicated by the X's and O's. The choice 10 of M iS made at the time of implementation and is dependent on the amount of smoothness desired: the larger the M the more smoothing applied to the decompressed video. The X's - and O's indicate which pixels are previously post-processed intensity pixels, X's, and which are decompressed intensity 15 pixels, O's, assuming a left to right movement across each row of the image frame. Other methods can also be used for moving through the decompressed image frame stored in memory, such as from right to left across each row, top to bottom or bottom to top across each column.
The general operation of the block diagram shown in FIG.
4 is the same as the block diagram in FIG. 3 and more detail is shown for one implementation of the combining circuit (318).
The combining circuit (318) comprises a predictor (402) and an 25 updater (404). The predictor (402) predicts the intensity of the current pixel. The updater (404) determines the post-processed pixel intensity.
The predictor (402) uses a one-dimensional non-causal 3 auto-regressive model of the local neighborhood of pixels (308). An adder (408) adds the first previous post-processed pixel (yb(n-1)) with the first decompressed pixel (x(n+l)) past the pixel under consideration (X(n)). The resultant of this operation is the sum of the current intensity levels (409) at 35 these two pixel locations. The coefficient (407) provided by wo 96114708 2 1 7 8 ~ 6 1 1 ~l/.Jb~S/lll87 first order linear prediction coefficient determiner (406) is then used to multiply (410) the sum of intensities (409) . This results in the estimate (412) of the original intensity at pixel location n. The estimate (412) is then compared (414) to the S decompressed pixel intensity (X(n)). The output of this comparison (415) is then multiplied (416) by the output (316) of the optimal filter gain deterrniner (314). The output of this value (417) is summed with the estimate of the intensity value (412), using an adder (418), resulting in the post-10 processed pixel intensity (320).
FIG. 5, numeral 500, is a diagram of a post-processing filter which suppresses mosquito and blocking artifacts and performs resizing in accordance with the present invention.
lS The post-processing filter comprises a memory unit (302), a signal variance determiner (310), an optimal filter gain determiner (314), and a combining circuit (318). The memory unit (302) receives and stores a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a 20 decompressed video sequence (216). The decompressed video sequence (216) is stored as a local neighborhood of pixels (308). As in FIG. 4, FIG. 5 reflects the case when M=2 which corresponds to the local neighborhood of pixels consisting of the five pixel elements indicated by the O's. The choice of M
25 is made at the time of implementation and is dependent on the amount of smoothness desired and the type of resize filter chosen: the larger the M the more smoothing applied to the decompressed video. The O's indicate which decompressed intensity pixels are to be used in the local neighborhood.
The mem~.ry unit (302) a'30 stores a predetermined plurality of inte. polation coefficients (501). The combining circuit (318) typically comprises a resize filter. The combining circuit accesses the memory (302) once. The 3 S combining circuit (318) comprises a filter weight determiner WO g6/14708 PCT/lJSg5/11187 21 7P,96 1 (502), a plurality of multipliers (504), and a summer (506).
The filter weight determiner (502) determines a plurality of filter coefficients (508). The plurality of multipliers (504) generates a plurality of weighted decompressed pixel 5 intensities (510). The summer provides the post-processed pixel intensity (320).
The post-processing filters (400 and 500) may be implemented using a Digital Signal Processor (DSP) such as the 1 0 MOTOROLA DSP56002 or an application specific integrated circuit (ASIC) and a Random Access Memory (RAM) unit. The rnemory unit (302) may be the RAM unit. The signal variance determiner (310), the optimal filter gain (314), and the predictor unit (402) may be implemented using the ASIC. The 15 updater may be implemented using two adders (414 and 418) and a multiplier (416) within the ASIC.
The present invention provides a method and apparatus for removing both mosquito and blocking artifacts from a 20 decompressed video sequence. With such a method and apparatus, it is possible to simultaneously resize the decompressed video while removing these artifacts. This method and apparatus has the following advantages over present post-processing filters: it is independent of signal-to-2 5 noise calculations, which are poor indicators of artifactlocations; it can perform resizing and post-processing with a single call to memory; a higher perceived viewing quality is obtained.
Although exemplary embodiments are described above, it will be obvious to those skilled in the art that many alterations and modifications may be made without departing from the invention. Accordingly, it is intended that all such alterations and modifications be included within the spirit and scope of the invention as defined in the appended claims.
The combining circuit (318) comprises a predictor (402) and an 25 updater (404). The predictor (402) predicts the intensity of the current pixel. The updater (404) determines the post-processed pixel intensity.
The predictor (402) uses a one-dimensional non-causal 3 auto-regressive model of the local neighborhood of pixels (308). An adder (408) adds the first previous post-processed pixel (yb(n-1)) with the first decompressed pixel (x(n+l)) past the pixel under consideration (X(n)). The resultant of this operation is the sum of the current intensity levels (409) at 35 these two pixel locations. The coefficient (407) provided by wo 96114708 2 1 7 8 ~ 6 1 1 ~l/.Jb~S/lll87 first order linear prediction coefficient determiner (406) is then used to multiply (410) the sum of intensities (409) . This results in the estimate (412) of the original intensity at pixel location n. The estimate (412) is then compared (414) to the S decompressed pixel intensity (X(n)). The output of this comparison (415) is then multiplied (416) by the output (316) of the optimal filter gain deterrniner (314). The output of this value (417) is summed with the estimate of the intensity value (412), using an adder (418), resulting in the post-10 processed pixel intensity (320).
FIG. 5, numeral 500, is a diagram of a post-processing filter which suppresses mosquito and blocking artifacts and performs resizing in accordance with the present invention.
lS The post-processing filter comprises a memory unit (302), a signal variance determiner (310), an optimal filter gain determiner (314), and a combining circuit (318). The memory unit (302) receives and stores a predetermined noise variance (304), a predetermined autocorrelation coefficient (306), and a 20 decompressed video sequence (216). The decompressed video sequence (216) is stored as a local neighborhood of pixels (308). As in FIG. 4, FIG. 5 reflects the case when M=2 which corresponds to the local neighborhood of pixels consisting of the five pixel elements indicated by the O's. The choice of M
25 is made at the time of implementation and is dependent on the amount of smoothness desired and the type of resize filter chosen: the larger the M the more smoothing applied to the decompressed video. The O's indicate which decompressed intensity pixels are to be used in the local neighborhood.
The mem~.ry unit (302) a'30 stores a predetermined plurality of inte. polation coefficients (501). The combining circuit (318) typically comprises a resize filter. The combining circuit accesses the memory (302) once. The 3 S combining circuit (318) comprises a filter weight determiner WO g6/14708 PCT/lJSg5/11187 21 7P,96 1 (502), a plurality of multipliers (504), and a summer (506).
The filter weight determiner (502) determines a plurality of filter coefficients (508). The plurality of multipliers (504) generates a plurality of weighted decompressed pixel 5 intensities (510). The summer provides the post-processed pixel intensity (320).
The post-processing filters (400 and 500) may be implemented using a Digital Signal Processor (DSP) such as the 1 0 MOTOROLA DSP56002 or an application specific integrated circuit (ASIC) and a Random Access Memory (RAM) unit. The rnemory unit (302) may be the RAM unit. The signal variance determiner (310), the optimal filter gain (314), and the predictor unit (402) may be implemented using the ASIC. The 15 updater may be implemented using two adders (414 and 418) and a multiplier (416) within the ASIC.
The present invention provides a method and apparatus for removing both mosquito and blocking artifacts from a 20 decompressed video sequence. With such a method and apparatus, it is possible to simultaneously resize the decompressed video while removing these artifacts. This method and apparatus has the following advantages over present post-processing filters: it is independent of signal-to-2 5 noise calculations, which are poor indicators of artifactlocations; it can perform resizing and post-processing with a single call to memory; a higher perceived viewing quality is obtained.
Although exemplary embodiments are described above, it will be obvious to those skilled in the art that many alterations and modifications may be made without departing from the invention. Accordingly, it is intended that all such alterations and modifications be included within the spirit and scope of the invention as defined in the appended claims.
Claims (7)
OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:
1. A method for obtaining a post-processed pixel intensity, for a frame in a decompressed video sequence residing in memory, using a filter which suppresses a visibility of a mosquito artifact and a blocking artifact based on a predetermined noise variance and a predetermined autocorrelation coefficient, comprising the steps of:
A) creating a local neighborhood of pixels, in a memory unit, based on the decompressed video sequence, wherein the local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of decompressed pixel intensities;
B) estimating a signal variance based on the local neighborhood of pixels;
C) calculating an optimal filter gain based on the signal variance, the predetermined noise variance, and the predetermined autocorrelation coefficient; and D) determining, using a summer, the post-processed pixel intensity based on the predetermined autacorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain, wherein the post-processed pixel intensity is further based on an estimate of an intensity of a selected pixel, wherein the estimate of the intensity of the selected pixel is a function of the predetermined autocorrelation coefficient, a decompressed pixel intensity of a next pixel relative to a predetermined scanning, and direction, and an immediately previously post-processed pixel intensity, wherein the estimate of the intensity of the selected pixel is determined by using an equation of the form:
where ~a (n) is the estimate of the intensity of the selected pixel, ~ is the predetermined autocorrelation coefficient, x(n+1) is a decompressed pixel intensity of a next pixel relatives to a predetermined scanning direction, and ~b (n-1) is an immediately previously post-processed pixel intensity.
A) creating a local neighborhood of pixels, in a memory unit, based on the decompressed video sequence, wherein the local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of decompressed pixel intensities;
B) estimating a signal variance based on the local neighborhood of pixels;
C) calculating an optimal filter gain based on the signal variance, the predetermined noise variance, and the predetermined autocorrelation coefficient; and D) determining, using a summer, the post-processed pixel intensity based on the predetermined autacorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain, wherein the post-processed pixel intensity is further based on an estimate of an intensity of a selected pixel, wherein the estimate of the intensity of the selected pixel is a function of the predetermined autocorrelation coefficient, a decompressed pixel intensity of a next pixel relative to a predetermined scanning, and direction, and an immediately previously post-processed pixel intensity, wherein the estimate of the intensity of the selected pixel is determined by using an equation of the form:
where ~a (n) is the estimate of the intensity of the selected pixel, ~ is the predetermined autocorrelation coefficient, x(n+1) is a decompressed pixel intensity of a next pixel relatives to a predetermined scanning direction, and ~b (n-1) is an immediately previously post-processed pixel intensity.
2. The method of claim 1, wherein the signal variance is estimated by using an equation of the form:
is an average intensity value in the local neighborhood of pixels, (2M+1) a neighborhood size, and is y(n) is a previously post-processed pixel intensity in the local neighborhood of pixels.
is an average intensity value in the local neighborhood of pixels, (2M+1) a neighborhood size, and is y(n) is a previously post-processed pixel intensity in the local neighborhood of pixels.
3. The method of claim 1, wherein the optimal filter gain is calculated by using an equation of the form:
and where .sigma.~ is the predetermined noise variance, .sigma.~ is the signal variance, and ~ is the predetermined autocorrelation coefficient.
and where .sigma.~ is the predetermined noise variance, .sigma.~ is the signal variance, and ~ is the predetermined autocorrelation coefficient.
4. The method of claim 1, wherein the post-processed pixel intensity is a resized post-processed pixel intensity based on a plurality of predetermined interpolation coefficients.
5. The method of claim 4, wherein memory is accessed one time for determining the resized post-processed pixel intensity.
6. A method for obtaining a post-processed pixel intensity, for a frame in a decompressed video sequence residing in memory, using a filter which suppresses a visibility of a mosquito artifact and a blocking artifact based on a predetermined noise variance and a predetermined autocorrelation coefficient, comprising the steps of:
A) creating a local neighborhood of pixels, in a memory unit, based on the decompressed video sequence, wherein the local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of decompressed pixel intensities;
B) estimating a signal variance based on the local neighborhood of pixels;
C) calculating an optimal filter gain based on the signal variance, the predetermined noise variance, and the predetermined autocorrelation coefficient; and D) determining, using a summer, the post-processed pixel intensity based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain, and wherein the post-processed pixel intensity is further based on an estimate of an intensity of a selected pixel, wherein the post-processed pixel intensity is determined by using an equation of the form:
~b(n)=~a(n)+C(n)~[x(n)-~a(n)]
where G(n) is the optimal filter gain, ~a (n) is the estimate of the intensity of a pixel, and x(n) is a decompressed pixel intensity.
A) creating a local neighborhood of pixels, in a memory unit, based on the decompressed video sequence, wherein the local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of decompressed pixel intensities;
B) estimating a signal variance based on the local neighborhood of pixels;
C) calculating an optimal filter gain based on the signal variance, the predetermined noise variance, and the predetermined autocorrelation coefficient; and D) determining, using a summer, the post-processed pixel intensity based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain, and wherein the post-processed pixel intensity is further based on an estimate of an intensity of a selected pixel, wherein the post-processed pixel intensity is determined by using an equation of the form:
~b(n)=~a(n)+C(n)~[x(n)-~a(n)]
where G(n) is the optimal filter gain, ~a (n) is the estimate of the intensity of a pixel, and x(n) is a decompressed pixel intensity.
7. A video compression system which includes a post-processing filter for suppressing a visibility of a mosquito artifact and a blocking artifact, the system comprising:
A) a video encoding unlit, operably coupled to receive a video sequence, for removing redundant information contained within the video sequence to generate a compressed video stream;
B) a channel encoding unit, operably coupled to the video encoding unit, for encoding the compressed video stream to generate an encoded compressed video stream which is set over a channel;
C) a channel decoding unit, operably coupled through the channel to the channel encoding unit, for decoding the encoded compressed video stream to generate the compressed video stream;
D) a video decoding unit, operably coupled to the channel decoding unit, for generating a decompressed video sequence; and E) a post-processing unit, operably coupled to the video decoding unit, for utilizing a post-processing filter, wherein the post-processing filter removes the mosquito artifact and blocking artifact from the decompressed video sequence based on a predetermined noise variance and a predetermined autocorrelation coefficient to provide a post-processed video sequence, wherein the predetermined noise variance is a value between 1 and 10, by creating a local neighborhood of pixels in a memory unit, based on the decompressed video sequence, wherein the local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of decompressed pixel intensities, estimating a signal variance based on the local neighborhood of pixels; calculating an optimal filter gain based on the signal variance, the predetermined noise variance, and the predetermined autocorrelation coefficient;
and determining, using a summer, the post-processed pixel intensity based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain, wherein the post-processed pixel intensity is further based on an estimate of an intensity of a selected pixel, wherein the estimate of the intensity of the selected pixel is a function of the predetermined autocorrelation coefficient, a decompressed pixel intensity of a next pixel relative to a predetermined scanning, and direction, and an immediately previously post-processed pixel intensity, and wherein the estimate of the intensity of the selected pixel is determined by using an equation of the form:
where ~a (n) is the estimate of the intensity of the selected pixel, ~ is the predetermined autocorrelation coefficient, x(n+1) is a decompressed pixel intensity of a next pixel relative to a predetermined scanning direction, and ~ (n-1) is an immediately previously post-processed pixel intensity.
A) a video encoding unlit, operably coupled to receive a video sequence, for removing redundant information contained within the video sequence to generate a compressed video stream;
B) a channel encoding unit, operably coupled to the video encoding unit, for encoding the compressed video stream to generate an encoded compressed video stream which is set over a channel;
C) a channel decoding unit, operably coupled through the channel to the channel encoding unit, for decoding the encoded compressed video stream to generate the compressed video stream;
D) a video decoding unit, operably coupled to the channel decoding unit, for generating a decompressed video sequence; and E) a post-processing unit, operably coupled to the video decoding unit, for utilizing a post-processing filter, wherein the post-processing filter removes the mosquito artifact and blocking artifact from the decompressed video sequence based on a predetermined noise variance and a predetermined autocorrelation coefficient to provide a post-processed video sequence, wherein the predetermined noise variance is a value between 1 and 10, by creating a local neighborhood of pixels in a memory unit, based on the decompressed video sequence, wherein the local neighborhood of pixels includes a plurality of previously post-processed pixel intensities and a plurality of decompressed pixel intensities, estimating a signal variance based on the local neighborhood of pixels; calculating an optimal filter gain based on the signal variance, the predetermined noise variance, and the predetermined autocorrelation coefficient;
and determining, using a summer, the post-processed pixel intensity based on the predetermined autocorrelation coefficient, the local neighborhood of pixels, and the optimal filter gain, wherein the post-processed pixel intensity is further based on an estimate of an intensity of a selected pixel, wherein the estimate of the intensity of the selected pixel is a function of the predetermined autocorrelation coefficient, a decompressed pixel intensity of a next pixel relative to a predetermined scanning, and direction, and an immediately previously post-processed pixel intensity, and wherein the estimate of the intensity of the selected pixel is determined by using an equation of the form:
where ~a (n) is the estimate of the intensity of the selected pixel, ~ is the predetermined autocorrelation coefficient, x(n+1) is a decompressed pixel intensity of a next pixel relative to a predetermined scanning direction, and ~ (n-1) is an immediately previously post-processed pixel intensity.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US08/334,718 US5802218A (en) | 1994-11-04 | 1994-11-04 | Method, post-processing filter, and video compression system for suppressing mosquito and blocking atrifacts |
| US08/334,718 | 1994-11-04 | ||
| PCT/US1995/011187 WO1996014708A1 (en) | 1994-11-04 | 1995-08-31 | Method, post-processing filter, and video compression system for suppressing mosquito and blocking artifacts |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CA2178961A1 CA2178961A1 (en) | 1996-05-17 |
| CA2178961C true CA2178961C (en) | 2000-05-23 |
Family
ID=23308507
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002178961A Expired - Fee Related CA2178961C (en) | 1994-11-04 | 1995-08-31 | Method, post-processing filter, and video compression system for suppressing mosquito and blocking artifacts |
Country Status (7)
| Country | Link |
|---|---|
| US (1) | US5802218A (en) |
| EP (1) | EP0738447B1 (en) |
| CN (1) | CN1138401A (en) |
| AU (1) | AU684232B2 (en) |
| CA (1) | CA2178961C (en) |
| DE (1) | DE69523565T2 (en) |
| WO (1) | WO1996014708A1 (en) |
Families Citing this family (53)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08237669A (en) * | 1995-02-28 | 1996-09-13 | Sony Corp | Picture signal processor, picture signal processing method and picture signal decoder |
| DE19509117C2 (en) * | 1995-03-17 | 1997-02-27 | Bosch Gmbh Robert | Procedure for monitoring the transmission quality of digitized signals |
| US5818973A (en) * | 1996-01-11 | 1998-10-06 | Umax Data Systems Inc. | Apparatus for digital correction in an image system |
| KR100200628B1 (en) * | 1996-09-30 | 1999-06-15 | 윤종용 | Image quality enhancement circuit and method thereof |
| KR100243225B1 (en) | 1997-07-16 | 2000-02-01 | 윤종용 | Signal adaptive filtering method for reducting blocking effect and ringing noise and filter thereof |
| GB2333657B (en) | 1998-01-22 | 2002-08-21 | Snell & Wilcox Ltd | Video signal compression |
| GB2333656B (en) * | 1998-01-22 | 2002-08-14 | British Broadcasting Corp | Compressed signals |
| GB2335104B (en) | 1998-03-06 | 2002-01-30 | British Broadcasting Corp | Cascading of up conversion and down conversion |
| GB2337389B (en) * | 1998-05-15 | 2002-05-29 | Snell & Wilcox Ltd | Video signal processing |
| US6108455A (en) * | 1998-05-29 | 2000-08-22 | Stmicroelectronics, Inc. | Non-linear image filter for filtering noise |
| US6327307B1 (en) * | 1998-08-07 | 2001-12-04 | Motorola, Inc. | Device, article of manufacture, method, memory, and computer-readable memory for removing video coding errors |
| GB9822087D0 (en) | 1998-10-09 | 1998-12-02 | Snell & Wilcox Ltd | Improvements in data compression |
| GB9822092D0 (en) | 1998-10-09 | 1998-12-02 | Snell & Wilcox Ltd | Analysis of compression decoded sequences |
| GB9822094D0 (en) | 1998-10-09 | 1998-12-02 | Snell & Wilcox Ltd | Improvements in data compression |
| GB9824061D0 (en) | 1998-11-03 | 1998-12-30 | Snell & Wilcox Ltd | Film sequence detection (nt4) |
| US6700623B1 (en) | 1998-12-10 | 2004-03-02 | Snell & Wilcox Limited | Video signal processing using triplets of pixels |
| DE60016533T2 (en) | 1999-09-29 | 2005-12-15 | Fuji Photo Film Co. Ltd., Minamiashigara | Device for detecting a fluorescence image |
| GB2361126B (en) * | 2000-04-05 | 2004-04-21 | Snell & Wilcox Ltd | Spatial video processing |
| JP4191884B2 (en) * | 2000-08-18 | 2008-12-03 | ジーイー・メディカル・システムズ・グローバル・テクノロジー・カンパニー・エルエルシー | Image processing method, image processing apparatus, and image photographing apparatus |
| US7054500B1 (en) * | 2000-12-06 | 2006-05-30 | Realnetworks, Inc. | Video compression and decompression system with postfilter to filter coding artifacts |
| US6993191B2 (en) * | 2001-05-04 | 2006-01-31 | Pts Corporation | Methods and apparatus for removing compression artifacts in video sequences |
| US7003173B2 (en) * | 2001-06-12 | 2006-02-21 | Sharp Laboratories Of America, Inc. | Filter for combined de-ringing and edge sharpening |
| US7003174B2 (en) * | 2001-07-02 | 2006-02-21 | Corel Corporation | Removal of block encoding artifacts |
| US6822675B2 (en) * | 2001-07-03 | 2004-11-23 | Koninklijke Philips Electronics N.V. | Method of measuring digital video quality |
| JP2004537731A (en) * | 2001-08-02 | 2004-12-16 | ザ リサーチ ファウンデーション オブ ステイト ユニヴァーシティ オブ ニューヨーク | Method and system for improving the solution of a system of linear equations |
| WO2003021936A2 (en) * | 2001-09-05 | 2003-03-13 | Emblaze Semi Conductor Ltd | Method for reducing blocking artifacts |
| FR2830307B1 (en) * | 2001-10-02 | 2003-12-05 | Eurocopter France | ROCKING POWER TRANSMISSION WITH PERIPHERAL FRONT TOOTHING WHEELS OF THE "FAR GEAR" TYPE |
| EP1303143A3 (en) * | 2001-10-16 | 2004-01-14 | Koninklijke Philips Electronics N.V. | Blocking detection method |
| US6993204B1 (en) | 2002-01-04 | 2006-01-31 | Pixon Llc | High speed signal enhancement using pixons |
| US7292733B2 (en) * | 2002-10-16 | 2007-11-06 | Matsushita Electric Industrial Co., Ltd. | Image processing apparatus and image processing method |
| US7471845B2 (en) * | 2004-01-06 | 2008-12-30 | Sharp Laboratories Of America, Inc. | De-ringing filter |
| US7430337B2 (en) * | 2004-01-06 | 2008-09-30 | Sharp Laboratories Of America, Inc. | System and method for removing ringing artifacts |
| US7697782B2 (en) | 2004-09-16 | 2010-04-13 | Sharp Laboratories Of America, Inc. | System for reducing ringing artifacts |
| US7620261B2 (en) * | 2004-11-23 | 2009-11-17 | Stmicroelectronics Asia Pacific Pte. Ltd. | Edge adaptive filtering system for reducing artifacts and method |
| US7778480B2 (en) * | 2004-11-23 | 2010-08-17 | Stmicroelectronics Asia Pacific Pte. Ltd. | Block filtering system for reducing artifacts and method |
| JP4774736B2 (en) * | 2004-12-27 | 2011-09-14 | カシオ計算機株式会社 | Image enlargement apparatus and imaging apparatus |
| US7680355B2 (en) * | 2005-05-02 | 2010-03-16 | Intel Corporation | Detection of artifacts resulting from image signal decompression |
| US7548660B2 (en) * | 2005-09-30 | 2009-06-16 | Intel Corporation | System and method of spatio-temporal edge-preserved filtering techniques to reduce ringing and mosquito noise of digital pictures |
| US7551795B2 (en) * | 2005-11-01 | 2009-06-23 | Samsung Electronics Co., Ltd. | Method and system for quantization artifact removal using super precision |
| US7869980B2 (en) * | 2005-11-03 | 2011-01-11 | International Business Machines Corporation | Using statistics to locate signals in noise |
| US7657113B2 (en) | 2005-12-21 | 2010-02-02 | Hong Kong Applied Science And Technology Research Institute Co., Ltd. | Auto-regressive method and filter for denoising images and videos |
| US8090210B2 (en) | 2006-03-30 | 2012-01-03 | Samsung Electronics Co., Ltd. | Recursive 3D super precision method for smoothly changing area |
| US7796834B2 (en) * | 2006-11-16 | 2010-09-14 | Texas Instruments Incorporated | Post-processing technique for noise reduction of DCT-based compressed images |
| US7925086B2 (en) * | 2007-01-18 | 2011-04-12 | Samsung Electronics Co, Ltd. | Method and system for adaptive quantization layer reduction in image processing applications |
| US8073276B2 (en) * | 2007-04-10 | 2011-12-06 | Vivante Corporation | De-ringing filter for decompressed video data |
| KR20100016272A (en) * | 2007-04-25 | 2010-02-12 | 픽슨 이매징 엘엘씨 | Image compression and decompression using the pixon method |
| US8305497B2 (en) * | 2007-07-27 | 2012-11-06 | Lsi Corporation | Joint mosquito and aliasing noise reduction in video signals |
| JP4967921B2 (en) * | 2007-08-10 | 2012-07-04 | セイコーエプソン株式会社 | Apparatus, method, and program for image processing |
| GB2460069A (en) * | 2008-05-15 | 2009-11-18 | Snell & Wilcox Ltd | Sampling conversion between formats in digital image processing |
| US9535993B2 (en) * | 2008-06-11 | 2017-01-03 | Enpulz, Llc | Creative work registry |
| US10123050B2 (en) * | 2008-07-11 | 2018-11-06 | Qualcomm Incorporated | Filtering video data using a plurality of filters |
| TW201021565A (en) * | 2008-11-24 | 2010-06-01 | Novatek Microelectronics Corp | Apparatus and method for motion adaptive interlacing |
| CN103096049A (en) | 2011-11-02 | 2013-05-08 | 华为技术有限公司 | Video processing method and system and associated equipment |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4794531A (en) * | 1984-11-07 | 1988-12-27 | Hitachi, Ltd | Unsharp masking for image enhancement |
| US4754492A (en) * | 1985-06-03 | 1988-06-28 | Picturetel Corporation | Method and system for adapting a digitized signal processing system for block processing with minimal blocking artifacts |
| FR2606187B1 (en) * | 1986-10-31 | 1988-12-09 | Thomson Grand Public | METHOD AND DEVICE FOR REDUCING THE NOISE GENERATED BY A DIGITAL IMAGE ENCODER-DECODER |
| US4761819A (en) * | 1987-02-27 | 1988-08-02 | Picker International, Inc. | Adaptive noise reduction filter for reconstructed images |
| US4941043A (en) * | 1988-06-14 | 1990-07-10 | Siemens Aktiengesellschaft | Method for reducing blocking artifacts in video scene coding with discrete cosine transformation (DCT) at a low data rate |
| JPH03129987A (en) * | 1989-10-14 | 1991-06-03 | Sony Corp | Method for coding video signal |
| WO1992012495A1 (en) * | 1990-12-31 | 1992-07-23 | E.I. Du Pont De Nemours And Company | Method of resizing an image, designing a filter therefor, and mapping the output image to the input image |
| DE69331719T2 (en) * | 1992-06-19 | 2002-10-24 | Agfa Gevaert Nv | Method and device for noise suppression |
| GB9321372D0 (en) * | 1993-10-15 | 1993-12-08 | Avt Communications Ltd | Video signal processing |
| US5512956A (en) * | 1994-02-04 | 1996-04-30 | At&T Corp. | Adaptive spatial-temporal postprocessing for low bit-rate coded image sequences |
| JP2673778B2 (en) * | 1994-02-22 | 1997-11-05 | 国際電信電話株式会社 | Noise reduction device for video decoding |
| US5561611A (en) * | 1994-10-04 | 1996-10-01 | Noran Instruments, Inc. | Method and apparatus for signal restoration without knowledge of the impulse response function of the signal acquisition system |
-
1994
- 1994-11-04 US US08/334,718 patent/US5802218A/en not_active Expired - Fee Related
-
1995
- 1995-08-31 AU AU35032/95A patent/AU684232B2/en not_active Ceased
- 1995-08-31 CN CN95191135A patent/CN1138401A/en active Pending
- 1995-08-31 EP EP95931692A patent/EP0738447B1/en not_active Expired - Lifetime
- 1995-08-31 WO PCT/US1995/011187 patent/WO1996014708A1/en active IP Right Grant
- 1995-08-31 CA CA002178961A patent/CA2178961C/en not_active Expired - Fee Related
- 1995-08-31 DE DE69523565T patent/DE69523565T2/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| US5802218A (en) | 1998-09-01 |
| EP0738447A4 (en) | 1998-02-04 |
| WO1996014708A1 (en) | 1996-05-17 |
| AU684232B2 (en) | 1997-12-04 |
| DE69523565T2 (en) | 2002-06-27 |
| CN1138401A (en) | 1996-12-18 |
| AU3503295A (en) | 1996-05-31 |
| EP0738447B1 (en) | 2001-10-31 |
| DE69523565D1 (en) | 2001-12-06 |
| CA2178961A1 (en) | 1996-05-17 |
| EP0738447A1 (en) | 1996-10-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2178961C (en) | Method, post-processing filter, and video compression system for suppressing mosquito and blocking artifacts | |
| Jeon et al. | Blocking artifacts reduction in image compression with block boundary discontinuity criterion | |
| US6983079B2 (en) | Reducing blocking and ringing artifacts in low-bit-rate coding | |
| Yang et al. | Removal of compression artifacts using projections onto convex sets and line process modeling | |
| US7526142B2 (en) | Enhancement of decompressed video | |
| JP2673778B2 (en) | Noise reduction device for video decoding | |
| US5818964A (en) | Method and apparatus for selecting an adaptive filter for image data | |
| US7860167B2 (en) | Apparatus and method for adaptive 3D artifact reducing for encoded image signal | |
| AU689048B2 (en) | Spatially adaptive filtering for video encoding | |
| EP3232664B1 (en) | Method for decoding using an interpolation filter | |
| JP4988715B2 (en) | Noise filtering method and apparatus in video coding | |
| CN100574433C (en) | Be used for reducing the method and apparatus of the pseudomorphism of video flowing | |
| JP2008508751A5 (en) | ||
| JP2002238049A (en) | Method for coding artifact reduction | |
| TW200535717A (en) | Directional video filters for locally adaptive spatial noise reduction | |
| EP1316206A1 (en) | Noise filtering an image sequence | |
| US7957469B2 (en) | Video comfort noise addition technique | |
| JP2002077629A (en) | Extent determining method of blocked artifacts in digital image | |
| EP0714210A2 (en) | Method of reducing mosquito noise generated during decoding process of image data and device for decoding image data using the same | |
| WO2000054220A1 (en) | Process, device and use for evaluating coded images | |
| Westen et al. | Adaptive spatial noise shaping for DCT based image compression | |
| JPH09130648A (en) | Moving image signal processor | |
| KR100282141B1 (en) | Space-Time Pre-Filter of Image Encoder | |
| JPH07336563A (en) | Image filtering method and image filtering device | |
| JPS61140291A (en) | Dynamic/static separator of animation picture signal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| MKLA | Lapsed |