JPH0691653B2 - Adaptive coding system - Google Patents

Description

DETAILED DESCRIPTION [Overview] By dividing a screen into small blocks and selecting inter-frame predictive coding, inter-filled predictive coding or intra-filled predictive coding according to the property of an input block signal, The coding efficiency is improved.

[Industrial application field]

The present invention relates to an image signal encoding method, and more particularly, to an improvement in an image signal encoding method for a signal such as an image signal of a TV conference system in which a still region is mostly moving and a large amount of movement occurs.

In the above coding method, it is desired that the coding be performed in accordance with the property of the input block signal and that the coding efficiency be good.

[Problems to be solved by conventional technology and invention]

An encoding method in which a conventional screen is divided into small blocks and predictive encoding is performed using the small blocks, and an interframe encoding method and a 1-field encoding method that uses a block at the same position as the encoding block on the previous frame plane as a prediction block. In the previous coding block, there is an intra-frame (inter-field) coding method that uses the closest block, but the former method has the same coding efficiency in the previous frame in the case of the still area, so the coding efficiency is good. However, in the moving area, since the same image is not present in the previous frame, the coding efficiency is lowered. In the latter method, in the case of the still area, the same one does not exist in the screen one field before, so in the still area. The coding efficiency is low, and in either case, there is a problem that high-efficiency coding cannot be performed in both the moving area and the still area.

[Means for solving problems]

The problem is that, as shown in the principle diagram of FIG. 1, the value of each pixel of the coding block (d) and the first shape having the same shape as the coding block at the same position as the coding block in the previous frame plane. Block (a)
Value of each pixel, the value of each pixel of a plurality of blocks (b) having the same shape as the coded block around the same position as the coded block on the front field surface, and the value of each pixel on the same field surface. The value of each pixel of the already-encoded block (c) having the same shape as the surrounding of the encoded block or the average value of the values of each pixel in the already-encoded block is input to the motion vector detector. A value obtained by combining the absolute value of the difference between the value of each pixel of the coding block (d) and the value of the pixel corresponding to each pixel of the first block (a), and the coding block A value obtained by combining the absolute value of the difference between the value of each pixel in (d) and the value of the pixel corresponding to each pixel in each of the plurality of blocks (b), and each pixel in the encoded block (d) Value of each of the pixels of the already-encoded block (c) According to the adaptive coding method of the present invention, the value of each corresponding pixel or the value obtained by summing the absolute values of the differences from the average value of each pixel is compared, and the smallest block is used as a prediction block. Will be resolved.

[Operation] The present invention uses the inter-frame coding method that uses the block (a) as a reference block in the still region, and when the motion is very large, the intra-frame prediction may be better. It is superior in encoding efficiency to use the encoded blocks around the encoding block as the predictive encoding, and in the case of the middle of the two, the inter-field encoding is superior in encoding efficiency. Focusing on that,
Encoding block (d) when encoding the block
Value of each pixel in the first frame (a) on the previous frame surface and the sum of the absolute values of the differences between the pixel values corresponding to the respective pixels in the first block (a), and A value obtained by summing the absolute values of the differences between the pixel value and the pixel value corresponding to each pixel of each of the plurality of blocks (b) on the front field surface, and the value of each pixel of the coding block (d) And a value obtained by summing the absolute values of the differences between the pixel values corresponding to the respective pixels of the already-encoded blocks (c) on the same field surface or the average value of the respective pixels, and it becomes the minimum. Since the block is used as the prediction block, the coding efficiency can be improved from the moving region to the still region.

〔Example〕

 FIG. 2 is a block diagram of an embodiment of the present invention.

In the figure, 1 is a subtractor, 2 is a quantizer, 3 is an inverse quantizer, 4 is an adder, 5 and 6 are filled memories, 7 is a block average value calculator, 8 is a 1 block line memory, and 9 is a dynamic memory. A vector detector, 10 is a variable delay circuit, and 11 is a delay circuit.

FIG. 2 shows an example in the case of intra-frame prediction in which one block (c) immediately above the current block is selected as the average value prediction as shown in FIG. 1 for real-time processing and hardware simplification. is there.

In the figure, the previous filled screen that is the output of the filled memory 6 and the previous frame screen that is the output of the filled memory 5, and the block average value calculator 7 and the one-line memory 8 are used for faster coding by one block line. The value obtained by calculating the average value of the block immediately above the block is input to the motion vector detector 9 and the variable delay circuit 10, and in the motion vector detector 9, the value of each pixel of the encoded block (d) A value obtained by summing the absolute values of the differences between the pixel values corresponding to the respective pixels in the first block (a) on the frame surface, the pixel values of the encoded block (d), and the pre-filled surface. Value of each pixel of the coding block (d), and the sum of the absolute values of the differences between the pixel values corresponding to the respective pixels of the plurality of blocks (b) in FIG. Block (c) The comparator compares the value of the sum of the absolute value of the difference between the average of the values or the pixels of the pixels respectively corresponding to the respective pixels of the people, select smallest block, and outputs as a prediction block.

Then, the subtracter 1 obtains the difference from the input block, and the quantizer 2 quantizes and sends out. The delay circuit 11 delays the input signal by the delay time until the prediction block is determined and output.

Since the motion vector detector 9 and the variable delay circuit 10 are described in Japanese Patent Application No. 59-44969, their description will be omitted.

At this time, rather than the motion vector detector 9,
Which of the average value of the previous frame screen and the block immediately above is detected, or the moved amount is collectively transmitted as motion vector information to the receiving side together with the prediction error quantized value output from the quantizer 2. .

For the reference block in the field, the mode value may be used instead of the average value. Further, not only the block immediately above but also the block diagonally above may be used.

Next, the third effect of the present invention will be described.
This will be described with reference to FIGS. 4 and 5.

FIG. 3 is a block diagram of the encoder. The difference from FIG. 2 is that a block validity / invalidity determination circuit 12 is added and a vector quantizer 13 is used as a quantizer.

Fig. 4 shows the case where one person's face is up (large amount of movement), and Fig. 5 shows the case of a conference scene (medium amount of movement) by three persons. One field encoding was performed.

In each figure, the curve (1) is for motion-compensated inter-frame coding + intra-frame average value prediction, the curve (2) is for inter-frame + intra-frame average value prediction, and the curve (3) is for simple frame. This is the case when going for a while.

The horizontal axis is entropy (bit / pel) and the vertical axis is S / N (db).

FIG. 6 is a diagram showing specifications.

In FIG. 4, the entropy becomes smaller as the compression rate becomes higher, but as can be seen by comparing the curves (2) and (3), if the intra-frame average value prediction is combined, the S / N is large even with the same entropy. The image quality is improved.

Since the same S / N can be achieved with a smaller entropy, the compression rate can be increased.

If motion compensation is further performed, S / N can be further improved and entropy can be reduced.

The same effect can be obtained when the amount of movement is medium as shown in FIG.

〔The invention's effect〕

In this way, in general, interframe coding will be used in the still region, intra-field prediction will be used when the motion is very large, and pre-interfield prediction will be used in the middle. , The coding efficiency can be improved from the moving area to the still area.

[Brief description of drawings]

 FIG. 1 is a diagram showing the principle of the present invention, FIG. 2 is a block diagram of an embodiment of the present invention, FIG. 3 is a block diagram of a specific encoder, and FIGS. 4 and 5 show the effects of the present invention. FIG. 6 is a diagram showing the specifications. In the figure, 1 is a subtractor, 2 is a quantizer, 3 is an inverse quantizer, 4 is an adder, 5 and 6 are filled memories, 7 is a block average value calculator, 8 is one block line memory, and 9 is A motion vector detector, 10 is a variable delay circuit, respectively.

Claims (1)

[Claims] 1. When a screen is divided into small blocks and predictive coding is performed in small block units, the value of each pixel of the coding block (d) and the value at the same position as the coding block on the previous frame plane. First block (a) having the same shape as the coding block
Value of each pixel, the value of each pixel of a plurality of blocks (b) having the same shape as the coded block around the same position as the coded block on the front field surface, and the value of each pixel on the same field surface. The value of each pixel of the already-encoded block (c) having the same shape as the surrounding of the encoded block or the average value of the values of each pixel in the already-encoded block is input to the motion vector detector. A value obtained by summing the absolute values of the differences between the value of each pixel of the encoded block (d) and the value of the pixel corresponding to each pixel of the first block (a), and the encoded block A value obtained by summing the absolute value of the difference between the value of each pixel of (d) and the value of the pixel corresponding to each pixel of each of the plurality of blocks (b) on the previous field surface, and the encoded block (d ), The value of each pixel and the coded block ( ) The value of the pixel corresponding to each of the respective pixels or the value obtained by summing the absolute values of the differences from the average value of the values of the respective pixels in the symbolized block is compared, and the minimum block is predicted. An adaptive coding method characterized by being used as a block.