WO2008101622A1

WO2008101622A1 - Device and method for setting a scan clock for broadband analog video signals

Info

Publication number: WO2008101622A1
Application number: PCT/EP2008/001085
Authority: WO
Inventors: Markus Waldner
Original assignee: Micronas Gmbh
Priority date: 2007-02-20
Filing date: 2008-02-13
Publication date: 2008-08-28
Also published as: EP2115729A1; DE102007008683A1

Abstract

The invention relate to a device for setting a scan clock (ta) for broadband analog video signals (Vin) having a pixel clock (tp) of a pixel frequency (fp), means (A) for analyzing the pixel clock (tp) and means (V) for changing the scan clocks (ta) with regard to the scan clock frequency (fa) and phase (phia), wherein the means (A) for analyzing are suited for detecting scanning line-related gradients (G) in the analog video signal (Vin), the gradients being produced by interference between the pixel clock frequency (fp) and the scan clock frequency (fa) if the analog video signal (Vin) is scanned at the incorrect frequency, and the location (O) of the gradients and for controlling the means (V) for changing such that, if gradients (G) are present, the scan clock (ta) is changed until no gradients (G) are detected any longer.

Description

description

Apparatus and method for setting a sampling clock for broadband analog video signals

The invention relates to a device and a method for setting a sampling clock for broadband analog video signals according to the preamble features of claims 1 and 7 respectively.

When processing video signals, the z. As generated by graphics cards, one is faced with the situation of having to work with many different image formats. An image format is characterized inter alia by the fact that it has a certain resolution, i. H. has an exact number of pixels (pixels) in the horizontal and vertical directions. A distinction is made between a total number of pixels and an active number of pixels. The total number of pixels also includes the areas of horizontal and vertical exchange gaps in which, among other things, synchronization information is transmitted. The active pixel count indicates the number of pixels that represents the pure image content.

Graphics cards, such as are common in personal computers, work time-discrete, ie they send one pixel after the other in sync with an internal clock. For example, an XGA standard image corresponds to 1024 active pixels horizontal and 768 pixels vertical. The total number of pixels is 1344 pixels in the horizontal direction and 806 pixels in the vertical direction. A complete image in the XGA standard consists of 1083264 pixels, so the graphics card needs 1083264 clocks to produce a whole a naloges XGA image to send. In a representation at a frequency of 60 frames per second, this means that 1083264 x 60 = 64995840 clocks per second are needed for transmission. This frequency is called the pixel clock frequency. If such an analog video signal is fed back into a digital system for video signal processing, it must be scanned in it at a frequency which corresponds to the pixel clock frequency. If a scan is performed with a frequency that deviates from the pixel clock frequency, so-called artefacts in the scanned image that correspond to a moire pattern, which results from the interference between the pixel clock frequency and the sampling clock frequency, ie the image is blurred in the appropriate places.

In order to be able to set the correct sampling clock with the appropriate correct sampling frequency during sampling, it has begun to standardize the image transmission. In such a standard, for. B. VESA standard, in addition to the active and the total image resolution other parameters such. B. the line frequency and the frame rate defined.

According to the prior art, such as. For example, US 6 522 365, US 709996 or US 6922188, various methods are known to measure the line and / or frame rate and / or the number of lines and based on this information on the sent format and thus on the Resolution close to the associated pixel clock or sampling frequency.

However, this known procedure has several weak points in practice. On the one hand, there are several different timing definitions in the VESA standard, for example, which result in different pixel clock frequencies and therefore different sampling frequencies to be set. For an XGA standard image with 1024 x 768 pixels at 60 Hz There are pixel clock frequencies that vary between 63.50 MHz and 64.99 MHz. On the other hand, it is noticeable that graphics card manufacturers do not always adhere to the exact format definitions prescribed in the VESA standard, and therefore the pixel clock deviates from the given frequency. Another weakness is that image formats that do not conform to a VESA standard can not be processed, ie that it is not possible to precisely set an unknown format that does not correspond to a standard. Another disadvantage of the previously known methods is that in the measurements that are made for format detection, the image content should remain as far as possible, which leads to significant problems with highly moving images.

The object of the invention is to provide an apparatus and a method for setting the sampling clock for broadband analog video signals, with which it is possible to precisely set the sampling clock frequency for an unknown format, which moreover does not have to correspond to a standard ,

This object is achieved by a device having the watch ^¬ paint of claim 1 and a method having the features of patent claim 7.

Advantageous developments of the invention are the subject of the dependent claims.

The basic idea of the invention is to use a moiré pattern, which results from sampling at a wrong frequency due to interference between the pixel clock frequency and the sampling clock frequency, for setting the sampling clock frequency and the phase of the sampling clock. If an analog video signal z. B. oversampled, then in addition to the plateaus of the pixels and the pixel transitions are sampled. In these transitions, the image is out of focus and results in a moiré fringe pattern. In order to detect these transitions, signal gradients are determined in the analog video signal, only having to determine whether a gradient exists or not. If the sampling clock deviates from the pixel clock by one clock period, a fuzzy column appears in the picture. If you z. B. 12 cycles over or undersampled, so you can see 12 blurred stripes.

Accordingly, there is claimed an apparatus for adjusting a sampling clock for broadband analog video signals having a pixel clock of a pixel clock frequency comprising means for analyzing the pixel clock and means for varying the sampling clock in frequency and phase, the means being suitable for analysis interfering with the image line which arise when the analog video signal is sampled at an incorrect frequency and control the means for varying such that the sampling clock is changed in the presence of interference until no interference is detected.

Preferably, the means for analysis are designed such that the interferences are detected on the basis of a gradient in the analog video signal at the sampling point.

Preferably, the means for analysis and the means for changing are designed such that in the presence of gradients, a first location of the gradients is detected, then the phase of the sampling clock is changed and then a second location of the gradients is detected. Based on the recorded places The gradient is then determined a direction of movement and the sampling clock is changed based on the determined direction of movement.

It is advantageous that, based on the determined direction of movement of the gradient and with the information about which direction the phase has been shifted, it can be determined whether an overscan or a subscan takes place. With a shift of the phase to the right and a present oversampling, the detected gradients also move to the right, ie. H. the detected second location of the gradients after the phase shift lies further to the right in the image line. If, on the other hand, there is an undersampling, the gradients travel to the left and the detected second location accordingly lies further to the left in the image line. With a phase shift to the left (reduction of the phase angle) the same applies inversely, d. H. in the case of over-sampling, the gradients move to the left, with subsampling to the right. To determine the gradients, for example, the method of German Patent Application No. 10 2005 025 453.5 "Device for determining a measure of a signal change and method for phase control" can be used.

In a preferred embodiment of the invention it is provided that the means for analyzing control means for changing such that in the presence of gradients and identical first location and second location of the gradient, the frequency remains constant and only the phase of the sampling clock is changed, until no more gradients are detected.

An advantage of this embodiment is that it is detected when the pixel clock frequency and the sampling frequency coincide. men and thus a faster convergence of the algorithm is ensured.

In a preferred development, if no gradients are detected, it is checked whether the analog video signal is present.

The advantage here is that it is ruled out that a correct sampling clock is erroneously assumed, although no image information is available, and therefore only no gradients are detected in the sampled signal.

In a further preferred embodiment, the means for analysis are designed in such a way that an image line is subdivided into a plurality of equally sized sections and the line-determined gradients are summed over an entire picture in a histogram according to their sections.

Ideally, there are the same number of sections as sampling clocks, but for reasons of hardware this is not always favorable, so that this embodiment also brings advantages. In particular, it is possible to set the beginning and the end of the entire section of a picture line to be measured, so that individual picture sections can also be enlarged. The fact that the line-related determined gradients are summed over an entire image and evaluated by means of a histogram further unambiguous information about the gradient distribution and thus about the shape of the interference pattern are obtained.

In a preferred embodiment of the invention, the means for analysis are partially or completely implemented as software. An embodiment of this form has the advantage that z. B. for histogram analysis or to determine the direction of movement of the gradient slightly different methods and algorithms can be used. It is Z. B. conceivable to determine the movement of the gradient by means of correlation Fourier transformation or by a cluster analysis. A determination of the position of the gradient by means of threshold or other methods is conceivable.

In a preferred embodiment of the invention, the means for changing the sampling clock as a phase-locked-loop (PLL) for adjusting the frequency and as a delay-locked-loop (DLL) for adjusting the phase are formed.

Such an implementation has the advantage that the frequency and the phase of the sampling clock can be set very precisely with PLL and DLL circuits.

According to the independent claim 8, a method is claimed for adjusting a sampling clock in frequency and phase for broadband analog video signals having a pixel clock of a pixel frequency. According to the method of the invention, the analog video signal is sampled at the sampling clock, image line-related signal gradients and their location in the analog video signal are determined and the sampling clock is varied in its sampling frequency and its phase until no more gradients are detected.

According to a preferred method of analysis, a first location of the gradients is detected in the analysis, then the phase of the sampling clock is changed, and then a second location of the sampling rate is changed. served. On the basis of the detected locations, a direction of movement of the gradients by means of which the sampling clock is changed is determined.

This method has the advantage already described above that, based on the direction of movement of the gradient with information about the direction of the phase shift, it is possible to determine whether an overscan or a subscan is present. This has the advantage that a targeted change of the sampling frequency is possible.

According to a further preferred method, the sampling clock frequency of the sampling clock is kept constant in the presence of gradients and identical first and second locations of the gradients, and only the phase of the sampling clock is changed until no more gradients are detected.

An advantage of this procedure is that it is detected when the correct sampling frequency is present and the optimum sampling point can be set via the phase adaptation.

According to a preferred embodiment of the method, a picture line is subdivided into a plurality of equally sized sections, and the gradients determined by line are summed over an entire picture in a histogram according to their sections.

An advantage of this procedure is that the summation over an entire image yields a clear statement about the distribution of the gradients.

The invention will be explained in more detail with reference to an embodiment with the aid of the accompanying figures. Show it: FIG. 1 shows a first block diagram of a circuit according to the invention,

FIG. 2 shows a second, more detailed block diagram of a circuit according to the invention,

FIG. 3 shows exemplary signal curves for the analog input signal, the sampled signal and the determined gradients and

FIG. 4 shows an example flow chart for a method according to the invention.

1 shows an analog video signal Vin with a pixel clock tp that has a scanning device 1. The sampling device 1 performs sampling at a sampling clock ta having a sampling clock frequency fa and a phase phia, and provides a sampled signal Va. The sampled signal Va can be supplied, for example, to a computer monitor which, for reasons of clarity, is not shown. Furthermore, in the sampling of the analog video signal Vin by an analyzer A, which determines whether in the analog video signal Vin Gradienten G are present at the sampling point, and image line-related determines at which location 0, these gradients G are. The analysis device A, in turn, controls a changing device V which is suitable for changing the sampling clock ta in its sampling clock frequency fa and its phase phia in accordance with the analysis. The block diagram shown in Figure 2 shows a further refinement of the block diagram of Figure 1. The analog video signal Vin is thereby fed to an analog-to-digital converter (ADC) by simultaneously a function for determining the gradient G is implemented. The analysis device A is divided according to Figure 2 in a hardware and a software part. The hardware part is a pixel-clock-recovery-PCR sensor to which the information obtained in the analog-to-digital converter ADC gradient G, a vertical synchronous signal Vsync and a horizontal synchronizing signal Hsync are supplied. The horizontal sync signal Hsync indicates the end of each image line, the vertical sync signal Vsync the end of an entire image. By means of the two synchronizing signals Vsync, Hsync, it is possible for the pixel-clock recovery sensor PCR to convert the information, determined by line, via the gradients G into a histogram, in which the information determined line by line is summed over an entire image. The histogram obtained by the pixel-clock-recovery-sensor PCR is made available to an analysis software which evaluates the histogram and accordingly controls the modifier V.

According to FIG. 2, the modifier V consists of a phase-locked loop, a so-called phase-locked loop PLL and a delay-coupled control loop, a so-called delay-locked-loop DLL. In this case, the phase-locked loop PLL can be fed with the vertical synchronizing signal Vsync, which marks the end of an entire picture. The phase-locked loop PLL is further controlled by the analysis software and serves to generate the sample clock frequency fa. The sampling clock frequency can be supplied to the delay-locked-loop DLL, which, controlled by the analysis software, sets the phase angle phia between the sampling clock ta and the pixel clock tp. By recursively traversing this control loop, an optimal setting of the sampling clock t a is achieved.

FIG. 3 shows exemplary signal profiles for the analog video signal Vin, the sampled video signal Va and the determined gradients G. In the case of the analog video signal Vin, a line with 640 pixels is shown, from which a sampled clock of 642 clocks per line produces a sampled signal Va. As reflected in the presentation of the scanned

Signal Va shows arise in two places, which can vary depending on the phase position, amplitude drops. The signal G shown in FIG. 3 shows the gradients G which can be determined from the analog video signal Vin. If the phase of the sampling clock is changed at the pixel clock, so change the

Amplitude drops and, accordingly, the detected gradients G their position in the examined line, and it can be concluded from this movement on whether an over- or a subsampling is present.

The flowchart shown in Figure 4 shows a possible implementation of the method according to the invention.

In a first step, the phase-locked-loop PLL, the delay-locked-loop DLL and the pixel-clock-recovery-sensor PCR are initialized. Subsequently, the sampling clock ta is set and, in a following step, the histogram which was generated by the pixel-clock-recovery-sensor PCR from the line-related ascertained data of the gradient G is read. If a new format is detected, the next step is to jump back to the initialization process. If no new format is available, the read histogram is examined for gradient clusters. If no gradient clusters are detected, This checks whether there is image data, and if so, jumps to the end of the process. If there are no image data, the previous process is run again from the reading of the histogram. In the event that gradient clusters are determined, the position of the gradient clusters is determined in a next step. Subsequently, the position of the gradient clusters from the currently read histogram is compared with the position of the gradient clusters from the histogram of the preceding measurement, and a partial direction of movement is determined. If, in the following step, a phase shift is determined which is smaller than one clock period, then in the following step the sampling clock ta is changed by a fraction of the sampling clock period and the process is run again from the time of reading the histogram. If the determined phase shift is greater than one clock period, the resulting direction of movement of the gradient clusters is determined in the following step. If a movement is detected, then in a following step the sampling frequency fa is adjusted as a function of the direction of movement and the search algorithm rhythm used and the sampling clock ta is set again. No movement is detected in the determination of the resultant direction of movement, the phase is changed phia the ex ^¬ tasttaktes ta until no more gradients are tektiert de-. If no more gradients detected, the process is completed.

LIST OF REFERENCE NUMBERS

scanning

A means of analysis, analysis device

G gradient

V means for changing, changing means

Vin analog video signal Va sampled video signal ta sampling clock fa sample clock frequency φa phase tp pixel clock fp pixel clock frequency

O place

ADC analog-to-digital converter

PLL phase-locked-loop

DLL delay-locked-loop Vsync vertical sync signal

Hsync horizontal sync signal

PCR pixel clock recovery sensor

R movement direction

Claims

claims

1. Apparatus for setting a sampling clock (ta) for broadband analog video signals (Vin) with a pixel clock (tp) of a pixel frequency (fp), means (A) for analyzing the pixel clock (tp) and means (V) for changing the sampling clock (ta) in sampling clock frequency (fa) and phase (phia) characterized in that the means (A) are suitable for analysis, when scanning the analog video signal Vin, image line related interference between the pixel clock frequency (fp) and the sampling clock frequency (fa) detect when the analog video signal (Vin) is sampled at a wrong frequency and control the means (V) for changing such that the sampling clock (ta) is changed in the presence of interference until no more interferences detected become.

2. Device according to claim 1, characterized in that the interferences due to a gradient in the analog video signal (Vin) are detected.

3. Device according to claim 2, characterized in that the means (A) for the analysis and the means (V) for the modification are designed in such a way that

- that in the presence of gradients (G) a first location (Ol) of the gradients (G) is detected, then the

Phase (phia) of the sampling clock (ta) is changed and then a second location (02) of the gradient (G) is detected,

- that on the basis of the recorded places (Ol, 02) a movement direction (R) of the gradient (G) is determined, and - that the sampling clock (ta) is changed on the basis of the determined direction of movement (R).

4. Device according to one of the preceding claims, characterized in that the means (A) for analysis control the means (V) for changing such that in the presence of gradients (G) and identical first location (Ol) and second location (02) the gradient (G), the sampling frequency (fa) remains constant and only the phase (phia) of the sampling clock (ta) is changed until no more gradients (G) are detected.

5. Device according to one of the preceding claims, characterized in that the means (A) are designed for analysis such that when no gradients (G) are detected, it is checked whether an analog video signal (Vin) is present.

6. Device according to one of the preceding claims, characterized in that the means (A) are designed for analysis such that a picture line is divided into several equal sections and the line-related determined gradient (G) over an entire picture in a histogram of their sections be summed up.

7. Device according to one of the preceding claims, characterized in that the means (A) for the analysis are partially or completely implemented as software.

8. Device according to one of the preceding claims, characterized in that the means for varying the sampling clock (ta) a phase-locked-loop (PLL) for setting the sampling frequency (fa) and a delay-locked-loop (DLL) to Adjusting the phase (phia) has.

9. A method of setting a sampling clock (ta) having a sampling frequency (fa) and a phase (phia) for wideband analog video signals (Vin) with a pixel clock (tp) of a pixel clock frequency (fp), s e n c i n e d d e r c h the steps

- sampling the analog video signal (Vin) with the sampling clock (ta) at the frequency (fa) and the phase (phia), - image line-related determination of gradients (G) and their location (O) in the analog video signal (Vin), and

- Changing the sampling clock (ta) in sampling frequency (fa) and phase (phia) until no more gradients (G) are detected.

10. The method of claim 9, wherein:

during the analysis, a first location (Ol) of the gradients (G) is detected, then the phase (phia) of the sampling clock (ta) is changed and then a second location (02) of the gradients (G) is detected,

- That on the basis of the detected locations (Ol, 02) a direction of movement (R) of the gradient (G) is determined, and

- That the sampling clock (ta) on the basis of the determined direction of movement (R) is changed.

11. The method according to claim 9 or 10, characterized in that in the presence of gradients (G) and identical first location (Ol) and second location (02) of the gradient (G), the sampling frequency (fa) remains constant and only the phase (phia) of the sampling clock (ta) is changed until none Gradients (G) are detected more.

12. Method according to one of claims 9 to 11, wherein a picture line is subdivided into several sections of equal size and the line-related determined gradients (G) are summed over a whole picture in a histogram according to their sections.