JPH02239787A - Picture coding control system - Google Patents

Abstract

PURPOSE:To realize coding control with less deterioration in coding picture quality by calculating a code error power, coding a relevant picture element or a prediction error of a picture element block with less information quantity when the power is large and coding the prediction error with large information quantity when the power is small through the control of coding. CONSTITUTION:A prediction signal VP selected by a prediction device 5 from a local decoding signal VL is subtracted by the input original signal VI at a subtractor 1 and the resulting signal is inputted to a coder 2 as a prediction error VE. A coded output signal VC is sent and a difference with an input original signal VI' via a variable delay device 3 simultaneously is calculated by a subtractor 7. The input original signal VI' is an original picture signal corresponding to the prediction signal VP and the calculated difference is a prediction signal error ep of the prediction signal VP. The prediction signal error ep is converted into an error power by a power calculator 8 and the coder 2 and the decoder 4 code the picture element with less information when the error power is large and when the value is small, the relevant picture element is coded with larger information quantity.

Description

[Detailed Description of the Invention] (Industrial Application Field) The present invention relates to a method of encoding an image signal, reducing the bit rate, and transmitting it digitally, such as in a video conference, a video telephone, and a digital television transmission. It is concerned with the method of predictively encoding image signals and encoding and transmitting prediction error signals.

(Conventional technology) Encoding video signals such as video calls and video conferences,
There are many image communication services that reduce the bit rate and transmit digitally. In such a video transmission method, in order to reduce the bit rate, when encoding the relevant pixel or pixel block, prediction is performed using pixels that have already been encoded and transmitted and decoded, and only the prediction error is calculated. Predictive coding for coded transmission is often used.

Predictive coding can be roughly divided into two types, depending on the position of the pixels used for prediction: intra-frame prediction, which uses decoded pixels that have already been coded and transmitted in the same frame, and intra-frame prediction, which uses already encoded pixels in the previous frame. There is inter-frame prediction in which prediction is performed using decoded pixels that have been transmitted as data, and it is known that the latter is more efficient for images with little movement. Furthermore, as a technique for further reducing the bit rate, motion compensated interframe prediction is also used, in which motion is detected from a video signal and pixels used for prediction are selected from the previous frame using this motion vector.

(Problems to be Solved by the Invention) However, when the transmission bit rate is extremely low, all of the significant information cannot be transmitted, resulting in deterioration of encoded image quality. In this state, the decoded pixels used for prediction contain large noise, so the prediction accuracy itself deteriorates, the power of the prediction error signal increases, and the amount of information required to encode it increases. . In such a case, what is encoded is almost all noise, so even though the amount of information required for encoding is large, the encoded image quality does not improve, and meaningless bits are consumed. There is a problem in that a vicious cycle occurs in which an effective amount of information cannot be allocated to other parts to be encoded.

The present invention is intended to solve the problems of the prior art described above, and is an image encoding control method that realizes encoding control with little deterioration in encoded image quality.

(Means for Solving the Problems) The first feature of the present invention is to compare the locally decoded pixels used for prediction of the corresponding pixel or pixel block with the original signal and calculate the encoding error power thereof, When this value is large, the prediction error of the corresponding pixel or pixel block is encoded with a smaller amount of information, and when this value is small, the encoding is restricted so that the prediction error is encoded with a larger amount of information. It's about doing. The second feature of the present invention is that in a method of orthogonal transformation coding of prediction errors in units of pixel blocks, the locally decoded pixels used for prediction of the corresponding pixel block are compared with the original signal, and the encoding error power is If this value is large, only the low-order transform coefficients of the prediction error of the corresponding pixel block are encoded and transmitted, and if this value is small, even the high-order transform coefficients are encoded. The purpose is to control the encoding for transmission.

(Principle of the invention) The principle of the invention will be explained using FIG. FIG. 1 is a block diagram of a predictive encoding device according to the present invention. Prediction signal ■P selected by predictor 5 from locally decoded signal VL
The subtracter 1 calculates the difference between the input original signal VI and inputs it to the encoder 2 as the prediction error VE, and the encoded output signal VC is encoded and transmitted. The difference with the signal VI' is also calculated by the subtracter 7. This input source signal VI' is an original image signal corresponding to the prediction signal VP, and the calculated difference value is the prediction signal error ep of the prediction signal ■P. The predicted signal error ep is converted into error power by the power calculator 8, which causes the encoder 2 and decoder 4 to encode the corresponding pixel with a smaller amount of information when the error power is large, and to encode the corresponding pixel with a smaller amount of information when this value is small. control is performed so that the corresponding pixel is encoded with a larger amount of information.

Further, the present invention will be explained as follows. Now, it is assumed that VI is the input original image signal to be encoded, P is the prediction signal selected from the locally decoded image signal, and V is the original image signal corresponding to the pixel used as the prediction signal. At this time, the encoding control signal Cn that controls the amount of encoding information is expressed by the following formula (1
) is defined by

Cn- (VP-VI ")" (1) As a means of controlling the amount of encoded information, the prediction error (VI-VP
) is often changed by changing the step width q for quantizing the quantizer, but if the present invention is applied to this quantizer control as an example, q may be controlled by the following equation (2).

q=f (Cn) (2) Here, f is an increasing function, an example of which is shown in FIG.

Now, in predictive encoding, the prediction error signal generated by the predictive encoder 9 as shown in FIG. Hybrid encoding has been widely used in recent years. In this, the invention is explained as follows.

Now, each pixel value of the pixel block of the input image signal to be encoded is {VI}, each pixel value of the pixel block of the locally decoded image signal used for prediction is {VP}, and the corresponding original image signal is Let each pixel value of a pixel block be {VI'}. At this time, the encoding control signal Cn is defined as shown in equation (1), and as shown in FIG.
i≦m, 1≦j≦n), 1≦i≦k, 1≦j≦1
The range R of coefficients for transmitting the coefficients is defined by the following equation (3).

R=f. (Cn) (3) Here, f+ is a function that determines the transmission range from the error power, and is defined such that the range R becomes narrower as the encoded control signal Cn becomes larger. That is, as the error power of the predicted signal increases, only lower-order transmission coefficients are transmitted, and higher-order coefficients (
(Even if the value is significant) it is truncated without being transmitted. How to determine the transmission range of coefficients? Although it may be possible, an example of f in the case of using an 8×8 pixel block is shown in FIG.

Here, R1 is the transmission coefficient range when C n < T h +,
R2 is the transmission coefficient range when C n < T h 2;
R3 is the transmission coefficient range when Cn<Th, and R4 is C
Transmission coefficient range when n < T h 4, R6 is C
Transmission coefficient range when n < T h s, R6 is C
Transmission coefficient range when n < T h a, Th+(1
≦i≦6) is a threshold value, 'rh+ <T h 2 <
T }1 s < T h 4 < T h s <
It is T he. As shown in the figure, in the present invention, a plurality of threshold values Th are provided for the encoded control signal Cn, and the encoded control signal Cn
The range of coefficients to be encoded is controlled to be changed depending on the size of the coefficient.

As described above, the present invention eliminates the consumption of meaningless bits that do not contribute to the encoded image quality by reducing the amount of information allocated to encoding when the error power of the predicted signal is large, and the By allocating it to a portion that contributes to improving encoded image quality, the overall encoded image quality can be significantly improved.

(Embodiment) FIG. 6 shows an embodiment of motion compensated hybrid coding according to the present invention, in which the portion surrounded by a dashed line is a characteristic portion added according to the present invention. The local decoded signal VL is written into the frame memory 24 for prediction, but at the same time, the difference between the local decoded signal VL and the original image signal v' delayed by the frame memory 15 is taken by the subtracter 23 and converted into power by the squarer 22. After being converted, it is input to the accumulator 2l, and one pixel block is added to the accumulator 2l.

(2) When the processing for the pixel block is completed, the contents of the accumulator 2l are stored in the error power memory 20. That is, the error power memory 20 stores the error power of the locally decoded image signal for one frame in units of pixel blocks. now,
A motion vector, MV, is detected by the motion amount detector 26,
When the predicted signal VP is selected by this, the error power corresponding to this is simultaneously read out from the error power memory 20, block boundary correction is performed by the corrector l7, and the relationship between the error power and the transmission coefficient range is written. The data is inputted into the memory 18 that contains the data. As a result, the transmission range pattern is output from the memory l8, the mask circuit 13 calculates the logical product of the transform coefficient and the transmission range pattern, and only the transform coefficients within the transmission range are input to the quantizer l4, and the encoded output signal VC is transmitted.

In addition, when the predicted portion Bp straddles the block boundary at an area ratio a:b+c:d as shown in FIG. 7, the block boundary correction is performed by
Corrected error power Ee is calculated from Ec and Ed using the following equation (4).

Ee = (aEa +bEb +cEc +dEd
)/(a+b+c+d) (4) (Effects of the Invention) As described above, the present invention compares locally decoded pixels used for prediction of the corresponding pixel or pixel block with the original signal, and calculates the encoding error power. is calculated, and when this value is large, the prediction error of the corresponding pixel or pixel block is encoded with a smaller amount of information, and when this value is small, the prediction error is encoded with a larger amount of information. By controlling the encoding, the consumption of meaningless bits that do not contribute to the encoded image quality is eliminated, and by allocating that bit to parts that contribute to improving the encoded image quality, the overall encoded image quality can be significantly improved. I can do it. Furthermore, the present invention has a configuration as shown in FIG.
This can be realized by simply adding a circuit for calculating the error power, a memory for storing the error power, and a storage means such as a ROM for storing the pattern of the coefficient transmission range.

Therefore, the present invention aims to significantly improve the encoded image quality of moving images in image communication services such as video conferences, video telephones, and digital television transmission in which moving image signals are encoded and digitally transmitted with a reduced bit rate. can be done, and the effect is great.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a predictive encoding device to which the present invention is applied to predictive encoding. FIG. 2 is a diagram showing an example of the relationship between the quantization step width q in encoding information amount control and the encoding control signal Cn that controls the encoding information amount. Third
The figure is a block diagram of hybrid encoding. FIG. 4 is a diagram showing an example of the range of coefficients to be transmitted in hybrid encoding. FIG. 5 is a diagram showing a coding control signal Cn and a transmission range in hybrid coding using an 8×8 pixel block as an example. FIG. 6 is a block diagram of a device when the present invention is applied to motion compensated hybrid coding. FIG. 7 is a diagram showing an example of block boundary correction according to the present invention in motion compensated hybrid coding. DESCRIPTION OF SYMBOLS 1... Subtractor, 2... Encoder, 3... Variable delay device, 4... Decoder, 5... Predictor, 6... Subtractor,
7... Subtractor, 8... Power calculator, 9... Prediction encoder, lO... Orthogonal transform encoder, l1... Subtractor, l2... Orthogonal transformer, 13. ...Mask circuit, 14
...Quantizer, l5...Frame memory, 16...
・Inverse quantizer, l7...corrector, l8...memory,
19... Inverse orthogonal transformer, 20... Error power memory,
2l...Accumulator, 22...Squaring device, 23...Subtractor, 24...Frame memory, 25...Subtractor, 2
6...Motion amount detector.

Claims (3)

[Claims] (1) In a method that predictively encodes a digital image signal and encodes and transmits the prediction error, the locally decoded pixels used for predicting the corresponding pixel or pixel block are compared with the original signal to calculate the encoding error power. and comparing the calculated value of the encoding error power with one or more predetermined thresholds, and controlling the encoding information amount of the prediction error to be encoded according to the threshold. Image encoding control method. (2) The image encoding control system according to claim 1, wherein the prediction error is encoded with a smaller amount of information as the value of the encoding error power increases. (3) In a method of predictively encoding a digital image signal and orthogonally transform encoding the prediction error, as the value of the encoding error power increases, only the lower-order transform coefficients of the prediction error transform coefficients are encoded. 2. The image encoding control method according to claim 1, wherein: the image encoding control method encodes the coefficients up to a higher order when the error power is small.