JPH0525435B2 - - Google Patents

Description

[Detailed description of the invention] (Industrial application field) The present invention relates to a technique for encoding moving image signals.

(Prior art) Conventionally, as shown in Fig. 2, a motion/static separation circuit performs motion/static separation on an input moving image signal, and signals indicating a motion region and a still region are supplied to an interframe encoder, and the interframe When the encoder generates a large amount of information and needs to reduce it, one method is subsampling, which thins out and encodes pixels for moving areas, and normal sampling, which does not thin out pixels, for static areas. was being carried out.

(Problem to be solved by the invention) Conventionally, subsampling was performed only for moving regions, but in reality, the amount of information generated in stationary regions is sometimes equivalent to the amount of information generated in moving regions. It may be a degree. In addition, due to the setting of the motion/still determination threshold in the boundary between a moving area and a still area or in the background area during panning, even though the area should be determined to be moving, it may be determined to be stationary, and subsampling is not applied. When large-area motion occurs, excessive information is generated, and coding stop is sometimes used as a means of coding control in the worst case.

(Means for Solving the Problems) According to the present invention, when encoding a moving image signal using inter-screen correlation, a moving region is usually detected and subsampling is applied only to the moving region. However, when the amount of occupancy of the moving region is larger than a predetermined threshold value, a coding control method for a moving image signal is obtained, which is characterized in that subsampling is applied to the entire screen.

Further, according to the present invention, there is provided an apparatus for encoding a moving image signal using inter-screen correlation, comprising: a motion/static determination means for separately determining a motion region and a still region; means for calculating and comparing whether or not the occupied amount is larger than a predetermined threshold; means for determining an application area for subsampling using the comparison result and the output of the motion/static determination means; and an input moving image. means for generating a difference signal from the signal and the prediction signal; quantization means for quantizing the difference signal; means for reproducing a first local decoded signal from the output of the quantization means and the prediction signal; means for reproducing a decoded signal; means for generating a second locally decoded signal by interpolating pixels that were not selected to be encoded in the subsampling using the first locally decoded signal; and the second locally decoded signal. Coding of a moving image signal, comprising: means for generating the predicted signal based on the signal; and means for variable length encoding a difference signal for a pixel selected to be encoded using the output of the quantization means. A device is obtained.

(Function) In the present invention, as shown in FIG. 1, a moving image signal is first separated into a moving area and a still area by a moving/static separation circuit. For example, as described in Japanese Patent Application No. 59-194110, ``A device for separating dynamic and static moving image signals,'' a screen is divided into blocks of a certain size, and the absolute value of the frame difference value of each pixel in each block is calculated within the block. There is a method in which the block is determined by adding the sum and comparing the result of the addition with a predetermined threshold value. Alternatively, a method called the gradient method can be used to obtain a pixel-by-pixel motion vector from the luminance gradient within a frame and the frame difference value, and a set of pixels for which this motion vector is not zero can be defined as a motion region. . The method of separating motion and static according to the present invention may be any method. The dynamic/static separation circuit outputs 1 when the area is in motion, and outputs 0 when it is in the stationary area, for example. Next, the occupancy amount judgment circuit calculates the proportion of the moving area within one screen based on the results of the motion and static judgment, and if it exceeds a certain threshold, it outputs 1 for that screen, and if it is below the threshold, it outputs 1. Outputs 0. For example, frame A~
If the amount of occupancy within the screen of the moving area in frame K is as shown in Fig. 3, then frame A,
For B, I, J, and K, the occupancy of the moving area is smaller than the threshold, so the occupancy determination circuit outputs 0. In frames C to H, since the threshold value is exceeded, 1 is output. The output of the dynamic/static separator circuit is delayed and adjusted so that it is in phase with the output of the occupancy determination circuit, and then sent to the OR circuit. In the case of frames A, B, I, J, and K where the output of the occupancy determination circuit is 0 and the proportion of the motion area is not so large, the OR circuit outputs the delay-adjusted motion and static data. The output of the separation circuit is output as is, and A, B, I,
This results in dynamic areas (shaded areas) such as J and K. When the output of the occupancy determination circuit is 1, that is, the proportion occupied by the moving area is large, the output of the OR circuit is 1 for frames C to H in FIG. (shaded area). The output of the OR circuit is then supplied to the interframe encoder, and when the output of the OR circuit is 1, that is, the moving area, the interframe encoder is a subsystem that thins out the black pixels in the diagonally shaded area in Figure 5. Sampling is performed, and when the output of the OR circuit is 0, it is a static region, so normal encoding without thinning is performed. When the proportion of moving areas is large, subsampling is performed on the entire screen to suppress excessive information that occurs at the boundaries between large moving areas and static areas, or in the background during panning, making it easier to encode. Encoding that can reproduce smooth motion without discontinuous motion due to stops can be performed.

(Example) An example of the present invention will be described in detail with reference to the drawings. FIG. 6 shows the encoder of the present invention. An input moving image signal is supplied to a timing circuit 1, a motion/static separation circuit 2, and a delay circuit 11 via a line 101. The timing circuit 1 detects a frame synchronization signal from the input moving image signal and supplies it to the occupancy determination circuit 4 via a line 102. Dynamic and static separation circuit 2
separates a moving region and a still region from an input moving image signal, outputs 1 for a portion determined to be a moving region, and outputs 0 for a portion determined to be a still region. The output of the dynamic/static separation circuit 2 is supplied to the occupancy determination circuit 4 and the delay circuit 12 via a line 201. The occupancy determination circuit 4 calculates the occupancy of the moving area on a frame-by-frame basis based on the frame synchronization signal supplied from the timing circuit 1, and for frames in which the occupancy exceeds a certain predetermined threshold,
It outputs 1 during that frame, and outputs 0 for frames whose occupancy is smaller than the threshold. The output of the occupancy determination circuit 4 is supplied to the OR circuit 5 via a line 401. The occupancy determination circuit 4 will be described in detail later. The delay circuit 12 delays the dynamic/static separation signal supplied from the dynamic/static separation circuit 2, matches the phase with the output of the occupancy determination circuit 4, and supplies the signal to the OR circuit 5. If the determination signal supplied from the occupancy determination circuit 4 is 1 in a certain frame, the OR circuit 5 determines that whatever the motion/static separation signal supplied from the delay circuit 12 is, it will be 1, that is, the frame will be in the entire motion area. Replace it and output it as a motion/static determination result. When the output of the occupancy determination circuit 4 is 0, the motion/static separation signal supplied from the delay circuit 12 is directly output as the motion/static determination result. The output of the OR circuit 5 is connected to the switch 7 via a line 57.
The signal is supplied to a variable length encoder 10 and an interpolation circuit 16. The delay circuit 11 delays the input moving image signal by an amount corresponding to the time required for separating motion and static and determining the amount of occupancy, and supplies the signal to the subtracter 3 . The subtracter 3 subtracts the moving picture signal supplied from the delay circuit 11 and the prediction signal supplied from the frame memory 9 to generate a difference signal. The output of the subtracter 3 is supplied to a quantizer 6. The quantizer 6 quantizes the difference signal supplied from the subtracter 3 and supplies it to the switch 7. The switch 7 outputs the difference signal supplied from the quantizer 6 as it is when the output of the OR circuit 5 is 0, that is, in the static region. When the output of the OR circuit is 1 and is in the moving region, subsampling of the moving region is performed by replacing the output of the switch 7 with 0, for example, for pixels to be thinned out. The output of switch 7 is supplied to adder 8 and variable length encoder 10. Adder 8 adds the difference signal supplied from switch 7 and the prediction signal supplied from frame memory 9, and reproduces the first locally decoded signal. The output of the adder 8 is sent to the interpolation circuit 16 in response to the motion/static determination result supplied from the OR circuit 5.
After undergoing interpolation of the thinned out pixels, it is supplied to the frame memory 9 as a second locally decoded signal.
The interpolation circuit 16 can perform interpolation for the thinned out pixels in the area where the motion/static determination result is 1 using neighboring pixels such as front, back, top and bottom, or values of the previous frame. Frame memory 9 delays the locally decoded signal supplied from interpolation circuit 16 by one frame time and supplies it to subtracter 3 and adder 8 as a predicted signal. The variable-length encoder 10 converts the difference signal supplied from the switch 7 into 0 for only pixels encoded in subsampling for areas where the motion/static judgment result supplied from the OR circuit 5 is 1.
For the area, all pixels are code-converted using efficient variable-length codes such as Huffman codes, and the motion/static judgment results output from the OR circuit 5 are code-converted to determine the speed of the transmission path and the coding. The converted code is output to the transmission line while matching the speed. Regarding subsampling, a switch 7 is provided in FIG. 6, but for pixels to be thinned out, the interpolation circuit 16 performs interpolation using neighboring pixels or previous frame values, and the variable length encoder 10 performs interpolation using neighboring pixels or previous frame values. Since the thinned out pixels are not subject to code conversion, the switch 7 can essentially be removed.

Next, the decoder of the present invention will be explained using FIG. The variable length encoder 13 performs code inverse conversion on the code-converted variable length code supplied from the transmission line and the motion/static determination result while matching the speed of the transmission line and the speed of decoding, and sends the difference signal to the line 1314. The motion/static determination result is supplied to the interpolation circuit 17 via a line 1317. The adder 14 adds the difference signal supplied from the variable length encoder 13 and the prediction signal supplied from the frame memory 15 to generate a first decoded signal. The interpolation circuit 17 uses this first decoded signal to interpolate the pixels that have been thinned out using the motion and static determination results supplied from the variable length decoder 13, and generates a second decoded signal. The same method as the encoder is used for interpolation. Second
The decoded signal is supplied to the frame memory 15 and output as a decoder output.

Next, the occupancy determination circuit 4 will be explained using FIG. The counter 402 is the dynamic and static separation circuit 2
Then, the number of motion areas of the motion/static determination signal supplied via the line 201 is calculated. The counting is reset every frame by a frame synchronization signal supplied from timing circuit 1 via line 102. The output of counter 402 is in register 40
3. The register 403 stores the amount of occupancy of the moving area supplied from the counter 402 on the line 102.
The amount of occupancy of the moving area in the frame is captured at the timing of the frame synchronization signal supplied via the frame synchronization signal. That is, the value immediately before the counter 402 is reset is taken in. The output of register 403 is supplied to determiner 404. The determiner 404 determines the amount of occupancy of the motion area supplied from the register 403.
Memory). The occupied amount of the dynamic area supplied from the register 403 is input to the ROM address. The threshold value for making the determination is statistically investigated in advance, and 0 is written to the address corresponding to the occupied amount smaller than the threshold value, and 1 is written to the address corresponding to the occupied amount larger than the threshold value. By setting this value, when a value smaller than the threshold is input to the ROM,
0 is output from the ROM, and when a value larger than the threshold is input, a 1 is output from the ROM. In this way, the amount of occupancy of the moving area can be determined. The output of the determiner 404 is supplied to the OR circuit 5 via the line 401.

(Effects of the Invention) As explained in detail above, when the amount of occupancy of the moving area exceeds a certain threshold, the entire screen is subsampled by replacing the entire screen with the moving area. This realizes an encoding method that suppresses excessive information generated in the background portion during panning, eliminates discontinuous motion due to stopping of encoding, and reproduces smooth motion. When the present invention is put to practical use in this way, its effects are extremely large.

[Brief explanation of the drawing]

Figures 1, 3, 4, and 5 are diagrams explaining the principle of the present invention, Figure 2 is a diagram explaining the conventional method, and Figures 6, 7, and 8 are diagrams explaining the principle of the present invention. It is a figure explaining an example of. In the figure, 1...timing circuit, 2...dynamic/static separation circuit, 3...subtractor, 4...occupancy determination circuit, 5...OR circuit, 6...quantizer, 7...
Switch, 8... Adder, 9... Frame memory, 10... Variable length encoder, 11, 12... Delay circuit, 13... Variable length decoder, 14... Adder,
15... Frame memory, 16, 17... Interpolation circuit, 402... Counter, 403... Register, 404... Determiner.

Claims (1)

[Claims] 1. When encoding a moving image signal using inter-screen correlation, a moving region is usually detected and subsampling is applied only to the moving region. A coding control method for a moving image signal, characterized in that subsampling is applied to the entire screen when the value is larger than a certain predetermined threshold. 2. A device for encoding a moving image signal using inter-screen correlation, comprising a motion/static determining means for separately determining a moving region and a still region, calculating the amount of occupancy of the moving region from the output of the means, and determining the amount of occupancy. means for comparing whether or not the value is larger than a predetermined threshold; means for determining an application area for subsampling using the comparison result and the output of the motion/static determination means; means for generating a signal;
quantization means for quantizing the difference signal; means for reproducing a first locally decoded signal using the output of the quantization means and the prediction signal; means for generating a second locally decoded signal by interpolating pixels that were not selected, means for generating the prediction signal using the second locally decoded signal, and pixels selected to be encoded based on the output of the quantization means. 1. An encoding device for a moving image signal, comprising means for variable length encoding a difference signal for pixels. 3. The encoding control method for a moving image signal according to claim 1, characterized in that the proportion of the portion determined to be a moving region in the screen is used as the amount of occupancy of the moving region.