JPS61164390A - Adaptive forecast coding device of inter-frame and between frames of animation picture signal - Google Patents

Abstract

PURPOSE:To improve the coding efficiency by using an input signal to apply inter-frame forecast with respect to a predetermined still region and apply adaptive forecast with respect to an animation region regardless of the decision result of the adaptive forecast system. CONSTITUTION:A frame memory 10 retards an input animation picture signal by 1 frame time, a subtractor 11 calculates frame difference, an adder 12 calculates the total sum of frame difference absolute value of a picture element in a clock, a comparator 13 compares the sum with a prescribed threshold value and outputs the result as animation/still information. A subtractor 14 outputs a difference between the input signal and a forecast signal from a selector 17 to a quantizer 15 as a forecast error. A discriminator 20 decides the inter-frame forecast when the animation/still information from the comparator 13 represents the still region and the picture element forecast system when not, and the selector 17, according to the forecast system, selects one of plural forecast signals from a forecast circuit 18. A code converter 16 converts a quantized forecast error signal and the animation/still information into a proper code and outputs the result.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to digital transmission of image signals.
The present invention relates to a highly efficient band compression transmission technology using an adaptive predictive coding method for J image signals.

(Prior art and its problems) In predictive coding of moving image signals, interframe prediction is generally suitable for still images or still images containing slow motion, while intraframe prediction is suitable for still images containing intense motion. It is known that it is suitable for images.In addition, as a coding method that adaptively uses these two prediction methods, the frame There is a known method that examines which is more efficient, inter-prediction or intra-frame prediction, and then determines the prediction method for the pixel based on the results and predetermined rules (Shimama et al.: rTV Conference Image Signal Motion-compensated inter-frame and intra-frame adaptive predictive coding] Proceedings of the 1985 National Conference of the Institute of Electronics and Communication Engineers, 1277).

However, in such a method, the prediction method of a pixel is estimated based on the fact that the encoding method of the neighboring pixel is appropriate, and if this estimation is incorrect, the prediction efficiency may decrease. Inevitable. For example, in FIG. 5, an object A crosses the screen to the left, but intra-frame prediction is generally more efficient for a region B that newly appears as the background of A. At this time, the stationary area C
Interframe prediction is originally suitable for the boundary region between
In other words, when intra-frame prediction is suitable for pixels in area B, intra-frame prediction is still selected. In addition, when the motion vector found in the paper by Shimama et al. is a zero vector, there is a method of regarding this as a stationary region and selecting interframe prediction, but this method uses local decoding that includes quantization noise. A motion vector is determined using a signal. Therefore, whether this motion vector is zero or not does not accurately correspond to the actual motion and static, and therefore, it is not always possible to expect improvement in coding efficiency through adaptive prediction.

(Objective of the Invention) An object of the present invention is to eliminate the above-mentioned drawbacks, select appropriate interframe prediction especially for the static region, and provide a more efficient adaptive predictive coding method. .

(Structure of the Invention) According to the present invention, a moving image signal is supplied, a means for intra-frame predicting the moving image signal and outputting an intra-frame predicted signal, and a means for inter-frame predicting the moving image signal and outputting an inter-frame predicted signal. means for outputting the intra-frame predicted signal, means for storing whether the intra-frame prediction signal or the inter-frame prediction has been applied to the already encoded pixel, and the output of the storing means is used to output the intra-frame predicted signal. , a selection means for outputting one of the interframe prediction signals as a prediction signal, and means for predictively encoding the video signal using the output of the selection means, between and within frames of a moving picture signal. In the adaptive predictive coding device, there is provided a means for separating a moving region and a still region of the input video signal; and means for selecting a predicted signal and, when the separation result is a moving region, outputting a selection signal to the selection means based on the output of the storage means. , an intra-frame adaptive predictive coding device is obtained.

(Detailed Description of Configuration) In the present invention, first, an input image signal before predictive coding is applied is separated into a motion area and a still area. This separation method, for example, divides the screen into blocks of a certain size, adds the frame difference values of each pixel within the block, and compares the addition result with a predetermined threshold value. A method is used to determine the movement of the block.

That is, since the motion and static regions are separated using an input signal that does not include quantization noise, the separation accuracy is high and the influence of motion and static determination errors on coding efficiency can be eliminated.

An explanation will be added using FIG. When a certain pixel A is a pixel in a still area, as the prediction method for this pixel, the optimal interframe prediction method for the still area is selected. When it is pixel B in a moving area, normal inter-frame and intra-frame adaptive prediction is performed. For example, as described in the above-mentioned literature, it is determined whether inter-frame prediction or intra-frame prediction is actually suitable for multiple pixels C, D, E, and F in the vicinity of the pixel to be predicted, and this combination and a predetermined prediction method are used. The prediction method for pixel B is determined based on the rules. That is, the present invention performs optimal interframe prediction for a still area determined using an input signal, and performs adaptive prediction for a moving area, regardless of the decision result of the adaptive prediction method. .

(Example of the invention) Next, the present invention will be explained in detail using the drawings.

FIG. 2 is one embodiment of an encoder using the present invention. 1
The moving image signal input from 000 is supplied to the delay circuit 30 and also to the frame memory 10. Frame memory 10 delays the input signal by one frame time and outputs it to subtracter 11 . The subtracter 11 calculates a frame difference and outputs it to the intra-block adder 12. The intra-block adder 12 includes an absolute value calculator and a necessary amount of line memory.
The comparator 13 calculates the sum of the frame difference absolute values of pixels within the block and outputs this to the comparator 13.

The comparator 13 compares the magnitude with a predetermined threshold, and outputs the result as dynamic information to the code converter 16 and the determiner 20 as 1 if the value is greater, and 0 otherwise.

The delay circuit 30 delays the input signal by the time necessary for the above-mentioned motion/static determination, and outputs the delayed signal to the subtracter 14 . The subtractor 14 is
The difference between the input signal and the prediction signal supplied from the selector 17 is calculated, and the result is output to the quantizer 15 as a prediction error. Quantizer 15 quantizes the prediction error and outputs it to code converter 16 and adder 19. The adder 19 is
The quantized prediction error and the prediction signal supplied via line 1719 are added, and the result is output as a locally decoded signal to prediction circuit 18 and optimal prediction determination circuit 22. The prediction circuit 18 has, for example, a frame memory in the case of inter-frame prediction, and a line memory or register in the case of intra-frame prediction, and generates a plurality of prediction signals and supplies them to the selector 17 and the ll&appropriate prediction determination circuit 22. The optimal prediction determining circuit 22 compares the local decoded signal supplied from the adder 19 with a plurality of prediction signals, and if the prediction method that is closer to the local decoded signal, that is, the optimal prediction method is interframe prediction, If it is intra-frame prediction, 0 is output to the storage circuit 21. The storage circuit 21 stores the optimal prediction method for pixels in the vicinity of the pixel to be predicted as 1 and 0, and outputs this to the determiner 20. The determiner 20 performs interframe prediction when the dynamic information supplied from the comparator 13 represents a still area, and otherwise performs frame prediction on neighboring encoded pixels supplied from the storage circuit 21. The information indicating whether inter-frame prediction or intra-frame prediction is the optimal prediction method is converted using a predetermined rule to determine the prediction method for the pixel, and output to the selector 17. This determiner 20 can be easily realized using, for example, a ROM.

That is, the dynamic information supplied from line 1320 and the line 2120
Optimum prediction information for neighboring coded pixels supplied from 2017 is used as address input, and output is sent to selector 17 via line 2017 as prediction method selection information. FIG. 4 shows an example of logic used in this ROM. For example, M/S indicates motion and static information where 1 indicates motion and 0 indicates stationary, as C, E, F.
is the optimal prediction information for neighboring pixels C, E, and F when pixel B in Fig. 1 is the pixel to be encoded.
represents the prediction method for pixel B (1 is on the interframe side).

For example, in the case of M/S: 0, C=E=F=0, intra-frame prediction would have been optimal for neighboring pixels, but since it is a stationary area, S=1, that is, inter-frame prediction will be selected. . The selector 17 selects one of the plurality of prediction signals supplied from the prediction circuit 18 according to the prediction method selection information S described above, and outputs it to the subtracter 14 and the adder 19 .

The code converter 16 converts the quantized prediction error signal and dynamic information into an appropriate code (for example, a Huffman code), and outputs the code to the transmission path 2000.

FIG. 3 is an example of a decoder corresponding to the encoder described above.

The code inverse converter 23 inversely transforms the coded dynamic information and prediction error supplied from the line 2000, and outputs the dynamic information to the determiner 27 and the prediction error to the adder 24, respectively. The adder 24 adds the prediction signal supplied from the selector 25 and the prediction error, and outputs the result as a decoded signal to the output terminal 3000, the prediction circuit 26, and the optimal prediction determination circuit 29. Prediction circuit 2
6. Optimal prediction determining circuit 29, memory circuit 28, determiner 27
, the operation of selector 25 is the same as in the encoder described above.

(Effects of the Invention) As described above, according to the present invention, an adaptive prediction method is realized that can determine motion and static for an input signal that does not include quantization noise for prediction errors, and has a high probability of selecting the optimal prediction method.
The present invention is of great significance, as prediction adapted to the local characteristics of the input signal is performed, resulting in improved coding efficiency.

[Brief explanation of drawings]

FIG. 1 is a diagram for explaining the present invention in detail, FIG. 2 is a block diagram showing an embodiment of an encoder using the present invention, FIG. 3 is a block diagram showing an embodiment of the decoder, and FIG. 4 is a diagram showing the logic of the determiner 20 in FIG. 2 and the determiner 27 in FIG. 3, and FIG. 5 is a diagram illustrating a case where the conventional method is not suitable. In the figure, 10...Frame memories 11, 14...Subtractor 12...Intra-block adder 13...Comparator 15...Quantizer 16... ... code converter 17.25...
... Selectors 18, 26 ... Prediction circuit
19.24...Adder 20, 29...Determiner 21.28...Memory circuit 22, 29
...Optimum prediction determining circuit 23... Code converter 30... Delay circuit. Fig. 1 Stationary area ・ ・ ・ ・
・ ・ ・ ・ @
・ ・ ・ ・ ・Mount heat...：Su... 1 Dynamic area Figure 4 Figure 5

Claims (1)

[Claims] A moving image signal is supplied, a means for intra-frame predicting the moving image signal and outputting an intra-frame predicted signal, a means for inter-frame predicting the moving image signal and outputting an inter-frame predicted signal, and an already encoded A means for storing whether an intra-frame prediction signal or an inter-frame prediction is correct for a pixel, and predicting either the intra-frame prediction signal or the inter-frame prediction signal using the output of the storing means. In an inter-frame/intra-frame adaptive predictive coding device for a moving picture signal, the device includes a selection means for outputting a signal, and a means for predictively coding the moving picture signal using the output of the selection means. means for separating a moving region and a still region of a signal; and if the separation result is static, causing the selection means to select an interframe predicted signal regardless of the output of the storing means; and means for outputting a selection signal to the selection means based on the output of the storage means if the area is a region.