JPS60144088A - Adaptive coding system - Google Patents

Abstract

PURPOSE:To obtain high compressing rate by detecting an absolute value in the unit of blocks and detecting the absolute value of difference between the present picture element and the signal shifted by one picture element in the scanning direction so as to switch the coding method through the result of the comparison of the both. CONSTITUTION:An input picture signal is fed to a moving vector detection section 1, compared with a picture signal before one frame so as to generate a dynamic vector VopY representing the amount and direction of the movement. A variable delay section 3 changes the delay amount by being applied with the dynamic vector Vpot detected by the dynamic vector detection section 1, the picture signal before one frame stored in a frame memory 2 so as to generate a forecast signal. A subtractor 4 subtracts a signal of forecast value from an input picture signal and a difference signal is generated. The difference sigmal is applied to a deciding circuit 5, from which logic 1 is generated as decision information when the direct coding has much information amount and logical ''0'' is generated when the coding with fixed inter-frame forecast has much information amount and logical ''0'' is generated when the coding with fixed inter-frame forecast has much information amount.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention The present invention relates to a band compression processing device for image signals, and in particular to a high compression ratio by adaptively switching encoding rules for motion-compensated blocks. The present invention relates to adaptive encoding methods that can be used.

Conventional technology and problems Motion compensation encoding method subdivides one frame of screen into arbitrary blocks, and generates a motion vector indicating the amount and direction of movement between blocks in each frame, and an arbitrary block from the corresponding block in the previous frame. predicted based on the prediction function of
The image signal is transmitted based on the error (prediction error) between the predicted value of the block in the current frame.

Conventionally, intra-frame encoding in such motion-compensated encoding methods involves either directly encoding and transmitting a motion-compensated block, or fixedly performing intra-frame prediction and transmitting a quantized output. It was by either method. However, these two methods have advantages and disadvantages; depending on the degree of movement on the screen, there are cases in which the former method provides a better compression ratio, and cases in which the latter method provides a better compression ratio. Therefore, no matter which method was adopted, good results could not be obtained in all cases.

Purpose of the Invention The present invention attempts to solve the problems of the prior art, and its purpose is to directly encode and transmit motion-compensated blocks, and to fix intra-frame prediction in a fixed manner. By adaptively switching between encoding and transmitting the quantized output depending on the input signal,
The object of the present invention is to provide an encoding method that provides a high compression rate.

Structure of the Invention The adaptive encoding method of the present invention detects the absolute value for each block, and also detects the absolute value of the difference between the current pixel and a signal obtained by shifting this pixel by one pixel in the scanning direction, and calculates the magnitude of both. The encoding method is changed depending on the comparison result.

Embodiment of the Invention FIG. 1 shows the configuration of an embodiment of the adaptive encoding system of the present invention. In the figure, 1 is a motion vector detection unit;
2 is a frame memory, 3 is a variable delay unit, 4 is a subtracter, 5
is a judgment circuit, 6 is a selector, 7 is a delay circuit (D), 8
is a subtracter, 9 is a quantizer (Q), 10 is an adder, 11
is a delay circuit (D) and No. 12 is an adder.

In FIG. 1, an input image signal is transmitted to a motion vector detection unit 1.
is added to the frame memory 2 and compared with the image signal of the previous frame stored in the frame memory 2 to generate a motion vector Vopt indicating the amount and direction of motion. The variable delay unit 3 uses the motion vector VopL detected by the motion vector detection unit 1.
is added to change the amount of delay, and the image signal of the previous frame stored in the frame memory 2 is delayed to generate a predicted value signal. The subtracter 4 subtracts the predicted value signal from the input image signal for each pixel to generate a difference signal. This difference signal is added to the determination circuit 5, which determines which encoding should be performed for each block as described later, and the determination circuit 5 provides determination information indicating that the amount of information is better when directly encoded. is large, and generates ``θ'' when the amount of information is larger if fixed intra-frame prediction and encoding are performed. This judgment information is added to the selector 6, and thereby the selector The selection of signals for the predicted pseudo-random sequence in step 6 is performed.

On the other hand, the difference signal from the subtracter 4 is applied to the delay circuit 7, and is delayed by the time until the determination operation in the determination circuit 5 is completed.

This requires that the selector 6 has finished switching before the next subtraction in the subtracter 8 is performed, but since the judgment in the judgment circuit 5 is made in block units, the input of the subtracter 8 is This is because we need to meet up. The output of the delay circuit 7 is applied to a subtracter 8, and the predicted value signal from the selector 6 is subtracted therefrom to generate a difference signal.

' This difference signal is applied to a quantizer 9 and quantized to generate quantized error information.

Further, this quantized error information signal is added to the signal from the selector 6 in an adder 10.

When the determination information is "1" and the output of the delay circuit 11 is selected by the selector 6, the decoded signal output from the delay circuit 11 for one pixel before is output as a signal of the predicted value in the frame, and the determination information When is "0", a no-signal state is selected and output as a predicted value signal. As a result, when the determination information is "1", the subtracter 8 subtracts the predicted value signal within the frame from the delayed difference value signal of the delay circuit 7 to generate a difference value signal. , this signal is quantized by a quantizer 9 to generate quantized error information. On the other hand, when the determination information is "0", the subtracter 8
No predicted value is input to the quantizer 9, and the quantizer 9 directly quantizes the delayed difference value signal of the delay circuit 7 to generate quantized error information.

FIG. 2 shows an example of the configuration of the determination circuit in FIG. 1. In the figure, 21 is an absolute value circuit, 22 is an adder, and 2
3 is a flip-flop (FF), 24 is a gate, 25 is a subtracter, 26 is a one-pixel delay circuit (CD), 27 is an absolute value circuit, 28 is an adder, 29 is a flip-flop (FF)
, 30 is a gate, and 31 is a comparator.

In FIG. 2, the absolute value of the difference value signal from the subtracter 4 shown in FIG. 1 is determined in an absolute value circuit 21. In FIG. The output of the absolute value circuit 21 is added to the cumulative value signal from the flip-flop 23 in the adder 22, so that the flip-flop 23 generates an output that accumulates the current difference value within one block. At this time, the gate 24 is closed for each block by the reset signal, and the flip-flop 23 is cleared for each block.

The difference value signal is also applied to a subtracter 25, where the difference value of the previous pixel is subtracted, thereby generating a difference value signal when prediction is made from the pixel of the previous pixel. This signal is the absolute value circuit 2
7, its absolute value can be found. The output of the absolute value circuit 27 is added to the cumulative value signal from the flip-flop 29 in an adder 28, so that the flip-flop 29 generates an output that is accumulated within one block.

At this time, the gate 30 is closed for each block by the reset signal, and the flip-flop 29 is cleared for each block. The comparator 31 has both flip-flops 23,
29 are compared, and when the output of the flip-flop 23 is large, "1" is output as the determination information, and when the output of the flip-flop 29 is large, "θ" is output as the determination information.

In this way, in the adaptive encoding method shown in FIG.
Depending on the motion vector detected by comparing the previous image signal and the current image signal, the image signal of the previous frame is variably delayed to obtain a predicted value for each pixel, and this predicted value and the current image signal are By comparing and determining the difference between the current difference value accumulated for each block and the signal obtained by accumulating the difference value predicted from the information of the previous pixel for each block, the information Determine trends in quantity changes.

Next, after delaying the difference value signal by a time corresponding to the determination operation time of the determination circuit, the difference between this difference value and the predicted value for each frame is calculated, and this is quantized to generate error information. At this time, according to the judgment result of the change in the amount of information, if the cumulative value of the current difference value is larger than the cumulative value of the amount of change between the predicted value of the difference value from the information one pixel before and the current difference value, quantization is performed. The difference value is calculated by using the signal output obtained by adding the error information and the predicted value for each frame and delaying the decoded signal in the current frame by one pixel as the predicted value in one frame. Generate a predicted value for each frame to calculate the error information by subtracting it from the signal of , and use this to calculate the quantized error information. When the difference value is smaller than the cumulative amount of change between the predicted value and the current difference value, the difference value is directly quantized without subtracting the predicted value, and the quantized error information is included. . In this way, the adaptive encoding method of the present invention switches the encoding method depending on the input information, so
A high compression ratio can be obtained regardless of the state of the input signal.

According to the adaptive encoding method of the present invention as described in detail,
A signal obtained by accumulating the difference value between the image signal and the predicted value information for each block, and a cumulative value obtained by accumulating the amount of change for each adjacent pixel in the difference value for each block. Depending on the size, the difference value between the predicted value predicted within a frame based on the difference value and the difference value may be quantized and encoded, or the difference value may be directly quantized and encoded. Therefore, a high compression ratio can be obtained regardless of the state of the input signal.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of an embodiment of the adaptive encoding method of the present invention, and FIG. 2 is a diagram showing an example of the configuration of a determination circuit. t-e vector detection unit, 2-frame memory, 3-variable delay unit, 4--subtractor, 5--judgment circuit, 6--selector, 7--delay circuit (D) subtracter, 8--・Subtractor,
9-quantizer (Q), 10-adder, 11-delay circuit (D), 12-.-adder, 21--absolute value circuit, 2
2-.- Adder, 23- Flip-flop, 24.-
・Gate, 25-Subtractor, 26--1 pixel delay circuit, 27-Absolute value circuit, 28-Adder, 29-Flip-flop, 30--Gate, 31--Comparator

Claims (1)

[Claims] The image information is divided into blocks, and the image information of the corresponding blocks is compared for each frame to obtain motion vector information, and prediction is made from the block of the current frame and the corresponding block of the previous frame according to the motion vector. In a motion compensation encoding method that encodes an image signal using error information between the predicted value information and the predicted value information, a signal obtained by accumulating a difference value between the image signal and the predicted value information for each block; The difference between the predicted value predicted within a frame based on the difference value and the difference value, depending on the magnitude of the difference value with the cumulative value obtained by accumulating the amount of change for each adjacent pixel in the difference value for each block. quantize and encode the values, or
Alternatively, an adaptive encoding method characterized in that the difference value is directly quantized and encoded.