KR0152018B1 - Dpcm encoding method including forced refresh - Google Patents

Abstract

The present invention relates to a DPCM encoding method and a device including a forced refresh. The DPCM encoding method includes an intra encoding step of an input video signal, an inter encoding step of encoding an error signal between an input video signal and a video signal compensated for, and a screen divided into unit blocks of a predetermined size, among which a forced intra encoding is performed. Selecting a unit block so as to be uniformly distributed in the entire screen and determining a forced intra pattern, and selecting and outputting an intra-coded signal and an inter-coded signal according to the forced intra pattern. As a result, by sequentially refreshing a certain number of blocks evenly distributed throughout the screen, a uniform amount of data is generated for each screen to maintain a constant image quality, and to remove artifacts in which a band-like wave flows. And speed up the recovery of the screen.

Description

DPCM encoding method including forced refresh.

1 and 2 are block diagrams of digital encoding and decoding of a typical typical video signal, respectively.

3 is a block diagram of DPCM encoding including forced refresh in accordance with the present invention.

4 is a flowchart showing a process of determining intra / inter mode of each block.

5 is an exemplary diagram of a forced intra pattern.

6 is an explanatory diagram of a forced intra block size.

* Explanation of symbols for main parts of the drawings

21: inter coder 22: intra coder

23: forced intra pattern generator 24: data amount comparison unit

25: OR gate 26: switching unit

27: energy comparison

The present invention relates to a DPCM encoding method and apparatus for video data signals, and more particularly, to a DPCM encoding method and apparatus including forced refresh.

Recently, video and audio signals are digitally stored, transmitted, decoded, and reproduced, and are attracting attention as an extremely important technology in HDTV, HD-VTR, digital VTR, digital camcorder, multimedia, videophone, etc. As the standardization activity is active, the CCITT, CCIR, ISO, etc. are trying to determine the appropriate standard encoding method for each application.

1 and 2 illustrate typical encoding and decoding processes that are widely used for encoding moving images. In order to reduce redundancy of the video signal and reduce data volume, a combination of transform coding, DPCM (Differential PCM), vector quantization, variable length coding (VLC), and the like is commonly used as shown in FIG. The input video signal first undergoes a conversion encoding process for data compression in the screen. In the N × N transform (1), the screen is divided into small blocks of N × N (for example, N × 8 = N × N = 8 × 8). For each block, a discrete cosine transform (DCT) and a discrete sine are performed for each block. Transform), DFT (Discrete Fourier Transform), Walsh-Hadamard Transform, or the like, are performed. When energy is mainly concentrated in the low frequency region by the transformation, the transformation coefficients are converted into representative values to the extent that it is difficult to detect by the human eye through quantization (2), and VLC (3) based on the statistical characteristics of these representative values Reduce the amount of data further The coded signal is transmitted through the buffer 4, and the quantization step size q that changes depending on the state of the buffer 4 to prevent the overflow or underflow of the buffer 4 is applied to the quantization 2. The bit rate is adjusted to control, and the quantization step q is transmitted to the receiving end together with the encoded video signal.

On the other hand, in view of the large temporal correlation between screens, data compression efficiency can be further improved by encoding only error signals between screens. That is, in the case of a still picture, there is no change between screens, and even in the case of a moving picture, the difference signal becomes very small when the motion is estimated and compensated using the motion vector. This is a dynamic compensation DPCM encoded loop consisting of 5 to 10 components in FIG. The motion vector MV is obtained for each macroblock of a relatively large size and transmitted together with a video signal encoded in the form of additional information. 2 illustrates a process of decoding the transmitted encoded signal through an inverse process of encoding.

By the way, as a characteristic of DPCM, the current signal is obtained by adding the decoded error signal to the stored previous signal by the trailing device 17 of FIG. 2, so that once an error occurs, the reference signal properly recovers the original signal. If you fail to do so, you will not be able to recover indefinitely. For example, a screen that can be used as a reference signal can be stored when the receiver is powered on or changes channels. Even if there is a reference signal, the screen generated by the accumulation of errors inherent in DPCM can be stored. In order to prevent deterioration and to recover the screen change due to an error on the transmission channel as soon as possible, the original image should be transmitted in the PCM mode (INTRA mode) instead of the DPCM mode (INTER mode) at regular intervals. This is called refresh.

As a conventional refresh method for transferring an original image in the PCM mode, there are the following methods.

First, as a method of refreshing in units of screens, the INTRA morph is encoded and transmitted once every few frames. The disadvantage of this method is that in the intra mode, data is generated much more than in the inter mode, so it is difficult to allocate bits between screens to maintain the image quality uniformly.

Secondly, in order to overcome the disadvantages of the first method, only a part of the screen is forcibly refreshed every frame, but after several frames, the refresh area is circulated according to the elapse of the frame so that all parts of the screen can be refreshed. For example, the band-shaped sections in the horizontal or vertical direction are determined and refreshed sequentially. According to this method, the data generation amount is relatively uniform compared to the first method, making it easier to maintain the image quality.

However, according to the proposed HDTV plan of General Instruments Inc., which adopts this second method, it is seen that the refresh boundary moves in the horizontal direction in some images because the refresh region is vertically shaped. Looks on.

Accordingly, the present invention improves the second method of the above-described refreshing of a portion of the screen sequentially, thereby including a forced refresh to remove the artifacts that appear to be flowing by allowing the refresh region to be distributed evenly throughout the screen without being formed in a band shape. An object of the present invention is to provide a DPCM encoding method and an apparatus thereof.

In order to achieve the above object, according to an aspect of the present invention, in the DPCM encoding method of a video signal including a forced refresh, the intra encoding step of encoding the input video signal by intra mode, the video compensation with the input video signal An inter encoding step of encoding an error of the received video signal, dividing a screen into unit blocks having a predetermined size, and selecting a unit block to be forcibly intra-coded so as to be spatially and homogeneously distributed within the entire screen to determine a forced intra pattern. The DPCM encoding method includes selecting and outputting an intra-coded signal in the case of a unit block of a forced intra position according to the forced in-pattern and outputting an inter-coded signal in the case of non-forced intra position. Is provided.

The position of the forced intrablock of the current screen is shifted from the position of each forced intrablock of the previous screen by a predetermined number of blocks so that after a predetermined time, all the unit blocks can be forced to be sequentially encoded. When this elapses, the refresh cycles through the entire screen. In this case, in order for the forced intrablocks to be evenly distributed on the time axis as the screens are sequentially connected, the position of the forced intrablocks of the current screen is preferably located near the center of the forced intrablock positions of the previous screen.

In addition, when the block is not of the compulsory intra position, it is possible to make the mode selection more efficient by comparing the output data amounts of the intra coding and the inter coding with each other and outputting the encoded output signal of the encoding having a small data amount. have. However, instead of comparing the amount of output data of intra coding and inter coding with each other, each energy according to the correlation correlation in each screen for each signal to be intra coded and inter coded is compared with each other, so The same effect can be obtained by selecting the coding for the output signal and outputting the encoded signal.

In order to process the screen in real time, it is preferable to divide the screen into a plurality of zones and perform parallel processing. At this time, when the forced intra patterns of the respective zones serving as the unit of parallel processing are all the same, the hardware can be designed very easily.

In addition, in the case of changing the channel of the receiver or turning on a new power supply, in order to speed up the recovery speed of the screen reproduced by the broadcast compensation, it is preferable that the size of the unit block to be intra coded is as large as possible, that is, forced intra coding. It is preferable to make the size of the unit block to be multiplied by the size of the macroblock for motion estimation.

According to another field of the present invention for achieving the above object, there is provided a DPCM encoding apparatus that is particularly suitable for implementing the above-described DPCM encoding method.

Hereinafter, with reference to the accompanying drawings showing a preferred embodiment of the present invention will be described in more detail the present invention.

3 is a block for implementing DPCM encoding according to the present invention. As can be seen from this figure, in the DPCM encoding of the present invention, the intra-encoding unit 22 which outputs the PCM-encoded input video signal and outputs the video signal compensated with the input video signal according to the DPCM loop of FIG. It has an inter coder 21 for encoding the error. Then, the forced intra pattern generator 23 divides the screen into unit blocks having a certain size, selects unit blocks to be forcibly coded among them, and generates 1-bit data thereof. The generation of the forced intra pattern will be described later in detail. The data amount comparing unit 24 compares the amount of output data of the intra coding unit 22 and the inter coding unit 21 with each other and generates one bit data regarding the comparison result. The one bit data of these data amount comparator 24 and the one bit data of the compulsory intra pattern generator 23 are input to the OR gate 25, and the output signal of the OR gate 25 supplies the switching unit 26. To control. The switching unit 26 selectively outputs the output signal of the intra coding unit 22 or the output signal of the inter coding unit 21 according to the control of the OR gate 25 output signal.

4 is shown as a flowchart to facilitate understanding of the intra / inter mode decision process in FIG. 3 of performing refresh according to the present invention. As can be seen in this figure, since the position of the unit blocks to be refreshed by being forced intra coded in the frame by the forced intra pattern generation unit 23 is predetermined in encoding each frame, the block is the forced intra position. If is in the intra mode, the switching unit 26 outputs the output signal of the intra coding unit 22. If the block is not a forced intra position, the data amount generated by the intra coding unit 22 and the data amount generated by the inter coding unit 21 are compared with each other, as in the conventional method. The switching unit 26 is controlled so as to output the output signal of the one having the smaller amount of data in comparison with the smaller one. As described above, the energy comparing unit 27 compares each energy according to each screen correlation with respect to the signal to be encoded by the intra coding unit 22 and the inter coding unit 21 without comparing the amount of data actually generated. The results can be used as well. That is, it is possible to replace the data amount comparing unit 24 with the energy comparing unit 27.

Next, the pattern determination of the intra intra screen forced intra block performed by the forced intra pattern generation unit 23 will be described.

FIG. 5 shows an example of a forced intra pattern, and for convenience of explanation, a case where the number of pixels on the screen is 960 lines x 1408 pixels, as shown in the proposal of General Linnstrument Inc., one of the full HD digital HDTV proposals in the United States. It is displayed. Of course, the present invention has nothing to do with the format of the screen. Blocks marked with x in FIG. 5 indicate forced intrablocks of the current screen, and blocks marked with o indicate forced intrablocks of the previous screen. As illustrated in FIG. 5, in order to ensure that relatively uniform data is generated for each screen and artifacts caused by refreshing are not visible, some rules must be observed in determining the forced intra pattern.

1. The screen is divided into unit blocks of a certain size, and the unit blocks to be forcibly intra-coded are selected to be spatially and homogeneously distributed throughout the screen.

In the case of Fig. 5, the screen is divided into 12 x 44 unit blocks, and thus the size of each unit block is 80 lines x 32 pixels. The unit blocks to be co-encoded in adjacent lines are spaced apart from each other by 7 unit blocks, so that the unit blocks to be co- intra coded are spatially and homogeneously distributed throughout the screen (+7 rule).

2. In case of parallel processing by dividing the screen into multiple zones for real-time processing of the screen, the forced intra pattern of each zone, which is a unit of parallel processing, should be identical to facilitate the design of hardware.

In FIG. 5, the screen is divided into four zones having eleven unit blocks in the horizontal direction, and each of these zones is constituted with the same pattern as the forced intra pattern can be seen in this figure (+ Rule 11).

3. The position of the forced intrablock of the current screen is shifted by a predetermined number of blocks from the value of each forced intrablock of the previous screen, and after a predetermined time, all the blocks are determined to be forced intra-coded sequentially. When the screen passes, the refresh cycles through the entire screen. At this time, the forced intrablocks of the current screen are located near the center of the forced intrablock positions of the previous screen in order to distribute the forced intrablocks evenly on the time axis as the screens are sequentially connected.

As can be seen in FIG. 5, the position (x) of the forced intrablock of the current screen is shifted by 3 blocks from the position (○) of the forced intrablock of the previous screen (+3 rule). Accordingly, as the frame elapses, in this example, when 12 frames elapse, all blocks are refreshed. That is, it takes 12 frames for a picture to be completely reproduced by a forced intra, which is generally 0.3 to 0.4 seconds, and 12 frames correspond to 0.4 seconds for a 30 frame / frame video system. In addition, since the position (×) of the forced intrablock of the current screen is almost at the center of the forced intrablock positions (○) of the previous screen, the artifacts that appear to be flowing due to the uniform distribution of the forced intrablocks on the time axis are generated. It does not occur.

4. The size of the unit block to be forced intra code is made as large as possible to speed up the recovery speed of the screen reproduced by the broadcast compensation.

6 is a comparison of the size of the macroblock for motion estimation and the unit block for the forced intra. In the illustrated example, the forced intrablock is five times the macroblock for the dynamic estimation, and the macroblock includes a plurality of change blocks ( For example, 8 × 8). By increasing the size of the forced intrablock in this manner, when the channel of the receiver is changed or when a new power is supplied, the full screen can be completely recovered within the shortest time. That is, in the example of FIG. 6, when 12 frames have elapsed when the power is turned on or the channel is changed, the picture is completely recovered when it is ideal, but there is a residue left without a complete recovery. It can be removed in time.

A process of encoding an image signal by DPCM encoding according to the present invention having the configuration as described above will be described with reference to FIG.

The input video signal is output by being PCM coded by the intra encoder 22 and encoded by the inter encoder 21 as a difference signal from the video signal compensated according to the DPCM loop. Each of these output signals is compared with each other by the data amount comparing unit 24, and one bit data relating to the comparison result is generated, for example, when the data amount of the intra coding unit 22 is small, a signal of 1 is generated. In the reverse case, that is, when the data amount of the inter coder 21 is small, a signal of zero is generated. As described above, as indicated by the dotted line in FIG. 3, the first energy according to the intra-correlation of the input image signal to be encoded by the intra encoder 22 by the energy comparator 27 and the inter, The same result can be obtained even by comparing the second energy according to the intra-correlation of the error signal between the input video signal to be encoded by the encoder 21 and the video signal compensated. Thus, the energy comparing unit 27 generates a signal of 1 when the first energy is small and a signal of 0 when the second energy is small.

On the other hand, the forced intra pattern generator 23 determines the position of the block to be intra intra screen, for example, as shown in FIG. 5 and generates 1 bit data thereof. That is, for example, the data of 1 is output in the case of a block of forced intra position, and the data of 0 is output in the case of the block which is not.

The one bit data from the data amount comparator 24 and the one bit data of the forced intra pattern generator 23 are input to the OR gate 25, and when either of the two input signals is 1 in the OR gate 25, In other words, if it is intra information, the switching unit 26 is switched to output the coded signal by the intra coding unit 22. When both 1-bit data are 0, the OR gate 25 switches the switching unit 26 so that the signal from the inter-coding unit 21 is output. Therefore, when the block is the forced intra position on the forced intra pattern, the output signal of the intra coding unit 22 is naturally output. When the block is not the forced intra position, the data amount is determined by the data amount comparison or the energy comparison. Alternatively, the forced energy refresh of the screen and efficient mode selection are made possible by outputting a coded signal having a lower energy.

As described above, according to the present invention, by sequentially refreshing a certain number of blocks evenly distributed throughout the screen, a uniform amount of data is generated for each screen to maintain a constant image quality, and a stripe wave flows. It can remove the seeming artifacts and speed up the screen's recovery.

Claims (16)

A DPCM encoding method of a video signal including forced refresh, the intra encoding step of encoding an input video signal by intra mode. And an encoding step of encoding an error of the video signal which is complementarily compensated with the input video signal. Dividing the screen into unit blocks of a predetermined size, and selecting a unit block to be forced intra-coded among them to be spatially and homogeneously distributed throughout the screen to determine a forced intra pattern. And selecting and outputting an intra-coded signal in the case of the unit block of the forced intra position according to the forced intra pattern, and outputting an inter-coded signal in the case of not being the forced intra position. The method of claim 1, wherein the position of the forced intra block of the current screen is shifted by a predetermined number of blocks from each of the forced intra block positions of the previous screen so that all the unit blocks can be sequentially coded sequentially after a predetermined time. DPCM encoding method characterized in that. The DPCM encoding method according to claim 2, wherein the position of the forced intra block of the current screen is near the center of the positions of the forced intra block of the previous screen. The coded output signal according to any one of claims 1 to 3, wherein in the case of not being a block of a forced intra position, the encoded output signal of the encoded data having a small data amount is output by comparing each output data amount of the intra coding and the inter coding with each other. DPCM encoding method. (Correction) The energy according to any one of claims 1 to 3, wherein the energy is compared with each other according to the correlation in each screen for each signal to be inter-coded with the intra-coded, if it is not a block of a forced intra position. And a coded signal for a signal having few signals. The DPCM encoding method according to any one of claims 1 to 3, wherein the screen is processed in parallel by dividing a plurality of zones. 7. The DPCM encoding method of claim 6, wherein the forced intra patterns of the respective zones are the same. 4. The DPCM encoding method according to any one of claims 1 to 3, wherein the size of the unit block to be forced intra-coded is several times the size of the macroblock for dynamic estimation. An DPCM encoding apparatus of an image data signal, comprising: an intra encoding unit of an input video signal. And an inter encoder which encodes an error of the video signal which is complementary to the video signal. A forced intra-pattern generator for selecting unit blocks to be coded intra-coded among unit blocks having a predetermined size in a screen so as to be spatially and homogeneously distributed throughout the screen, and generating 1 bit data thereof. And a switching unit for selectively outputting an intra-coded signal when the unit block of the forced intra position is a unit block of the forced intra position according to the 1-bit data relating to the forced intra pattern. . The forced intra pattern generator determines that the position of the forced intra block of the current screen is shifted by a predetermined number of blocks from the position of the forced intra block of the previous screen so that all blocks can be sequentially encoded after a predetermined time. DPCM encoding apparatus characterized in that. 11. The DPCM encoding apparatus of claim 10, wherein the position of the forced intra block of the current screen is close to the center of the positions of the forced intra block of the previous screen. 12. The apparatus according to any one of claims 9 to 11, further comprising a data amount comparison unit for generating one bit data by comparing the output data amounts of the intra coding and inter coding with each other. And a switching control by an output signal of an OR gate receiving bit data and 1-bit data of the forced intra pattern generator. (Correction) The energy comparison part according to any one of claims 9 to 11, further comprising an energy comparison unit for generating one bit data by comparing each energy according to each screen correlation with respect to each signal to be inter-coded and inter-coded. And the switching unit is controlled to be controlled by an output signal of an OR gate that receives 1 bit data of the energy comparator and 1 bit data of the forced intra pattern generator. 12. The DPCM encoding apparatus according to any one of claims 9 to 11, wherein the DPCM encoding unit has a plurality of DPCM encoding units corresponding to the number of divided sections for parallel processing of each section of the plurality of equally divided screens. The DOCM encoding apparatus of claim 14, wherein the forced intra patterns of the respective zones are the same. 12. The DPCM encoding apparatus according to any one of claims 9 to 11, wherein the size of the unit block to be forced intra-coded is several times the size of the macroblock for dynamic estimation.