JPH0541863A - Visual telephone set - Google Patents

Abstract

PURPOSE:To prevent increase in generated code quantity due to miniaturization of hardware scale, low cost and pseudo matching by adding a discrimination circuit discriminating it that the motion compensation prediction is effective and detecting a motion vector only from a macro block requiring to detect the motion vector. CONSTITUTION:A signal of a memory A103 is inputted to a read discrimination circuit A104 for each macro block. The discrimination circuit A104 obtains a dynamic range for each macro block to compare it with a threshold level and when the dynamic range is less than the threshold level, a mode setting device 105 is set to the in-frame coding mode and when the dynamic range exceeds the threshold level, the mode setting device 105 is set to the inter-frame coding mode. The brightness of each picture element in the macro block is represented in 8-bit and 256-gradation and when the threshold level is set to 50 or below, the macro block with less brightness change and less harmonic component of the picture signal is subjected to in-frame coding.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video coding unit in a videophone.

[0002]

2. Description of the Related Art In a conventional videophone, all macroblocks are set in an information source encoder in an interframe coding mode with motion vector detection except when the intraframe coding mode is forcibly set. It was set and encoded.

[0003]

However, in the above-mentioned prior art, in the case of a macroblock having a small dynamic range, the accuracy of motion vector detection is poor and the generated code amount increases due to pseudo matching, or motion vector detection is performed in all macroblocks. Therefore, there is a problem that the scale of the hardware becomes large.

Therefore, the present invention solves such a problem, and an object thereof is to detect a motion vector of only a macroblock which needs a motion vector detection only by adding a simple determination circuit, and thereby to reduce the hardware scale. It is to prevent the increase of the generated code amount due to the pseudo matching of the motion vector detection in the case of a macroblock having a small dynamic range while realizing the miniaturization and the price reduction of the above.

[0005]

The present invention provides (1) a video input camera, a memory for recording one frame of a signal from the video input camera, and a C for encoding the signal from the memory.
CITT Recommendation H.264 In a videophone having a 261 compliant information source encoder, the signal of the memory is read for each macro block of 16 × 16 pixels, the dynamic range of each macro block is calculated and compared with a threshold value, and the dynamic range is calculated. Is less than or equal to the threshold value, the information source encoder is set to the intraframe coding mode, and when the dynamic range exceeds the threshold value, the determination circuit A is set to set the information source encoder to the interframe coding mode. Characterized by having,
(2) Video input camera, memory for recording one frame of the signal from the video input camera, CCITT Recommendation H.264 for encoding the signal from the memory. In a videophone equipped with a 261 compliant source encoder, the signal in the memory is
Read out for each macroblock of 16 pixels, find the dynamic range for each macroblock, and compare with the threshold,
If the dynamic range is less than the threshold value, the information source encoder is set to the inter-frame coding mode in which the motion vector is 0, and if the dynamic range exceeds the threshold value, the information source encoder is set to the inter-frame with motion vector detection. And a determination circuit B for setting the encoding mode,
(3) In the judgment circuit A or the judgment circuit B, 16 × 16
MSB side N of the luminance values of all pixels in the macroblock of pixels
If the bits (N = 4 or 5 or 6) have the same value, it is determined that the dynamic range is less than or equal to the threshold value, and the MSB side N bits (N = 4 or N) of the luminance values of all the pixels in the 16 × 16 pixel macroblock. If 5 or 6) are not the same value, it is determined that the dynamic range exceeds the threshold value.

[0006]

(Embodiment 1) FIG. 1 is a block diagram showing an embodiment of the present invention. The image signal input from the camera 101 is filtered by the pre-processing circuit 102 to obtain CIF or QC.
The format is converted to IF and stored in the memory A 103. The output of the memory A 103 is encoded by the information source encoder 115, output from the output 114, Huffman encoded, further video signal multiplexed, channel encoded, and transmitted. The above basic flow is based on CCITT Recommendation H.264. 261 is compliant. Next, the inside of the information source encoder 115 will be described. During intra-frame coding, the switch 106 and the switch 10
7 is connected to the upper side, and 8 × is provided by the discrete cosine converter 108.
Two-dimensional discrete cosine transform is performed in units of 8 pixels, and the transform result is quantized by the quantizer 109 and output from the output 114. The quantized signal is inversely transformed by an inverse quantizer 110 and an inverse discrete cosine transformer 111, and the motion compensation predictor 11
Store in 2. On the other hand, during interframe coding, the switches 106 and 107 are connected to the lower side, and the memory A1
The signal No. 03 is input to the motion compensation predictor 112, the motion vector is detected and the motion compensation prediction is performed with respect to the frame previously encoded, and the memory A is passed through the loop filter 113.
The difference from the signal of 103 is taken, and the discrete cosine converter 108 performs two-dimensional discrete cosine conversion in 8 × 8 pixel units, and the conversion result is quantized by the quantizer 109 and output from the output 114. The quantized signal is inversely transformed by the inverse quantizer 110 and the inverse discrete cosine transformer 111 and stored in the motion compensation predictor 112. The intra-frame coding mode and the inter-frame coding mode have been described above. In H.261, it is possible to switch between intraframe and interframe modes for each macroblock. In this embodiment, the signal of the memory A103 is read out for each macro block of 16 × 16 pixels and input to the determination circuit A104, the determination circuit A104 obtains the dynamic range of each macro block and compares it with a threshold value to determine the dynamic range. If is less than or equal to the threshold, the mode setter 105 is set to the intraframe coding mode, and if the dynamic range exceeds the threshold, the mode setter 105 is set to the interframe coding mode. The luminance value of each pixel in the macroblock is represented by 256 gradations of 8 bits, and when the threshold value is set to a value of 50 or less, the macroblock with less luminance change and less high-frequency components of the image signal is intraframe-coded. It will be. A macroblock having a small change in luminance and a high frequency component of an image signal has a small amount of generated code even if intra-frame coding is performed, and the effect of further reducing the generated code amount by inter-frame coding is small. Also, depending on the configuration of the motion vector detection circuit, the generated code amount often increases due to the interframe coding. This is because when a motion vector is detected in a macro block having a small dynamic range, pseudo matching is likely to occur due to the influence of noise and the like. Furthermore, in this method, the macroblock for intra-frame coding does not detect the motion vector, so that the overall throughput is increased. This is because the motion vector detection circuit has a search range of ± 1.
The sum of absolute differences of all pixels in a macroblock is calculated for each search in the range of 5 pixels, and the minimum distortion is calculated in the entire search range. This is because the overall throughput could not be increased because it took an extremely long time. However, in this method, the number of macroblocks for motion vector detection is reduced, so that the throughput is increased without changing other circuits.

(Embodiment 2) FIG. 2 is a block diagram showing an embodiment of the present invention. The image signal input from the camera 201 is filtered by the pre-processing circuit 202, and then CIF or Q
The format is converted to CIF and stored in the memory A 203. The output of the memory A203 is encoded by the information source encoder 215, output from the output 214, Huffman encoded, further video signal multiplexed, and channel encoded for transmission. The above basic flow is based on CCITT Recommendation H.264. 261 is compliant. Next, the inside of the information source encoder 215 will be described. At the time of intraframe coding, the switch 206 and the switch 20
7 is connected to the upper side, and 8 × is provided by the discrete cosine converter 208.
Two-dimensional discrete cosine transform is performed in units of 8 pixels, and the transform result is quantized by the quantizer 209 and output from the output 214. The quantized signal is inversely transformed by the inverse quantizer 210 and the inverse discrete cosine transformer 211, and the motion compensation predictor 21
Store in 2. On the other hand, during interframe coding, the switches 206 and 207 are connected to the lower side, and the memory A2
The signal No. 03 is input to the motion compensation predictor 212, the motion vector is detected between the frame previously encoded and the motion compensation prediction is performed, and the motion compensation prediction is performed through the loop filter 213.
The difference from the signal of 203 is taken, the two-dimensional discrete cosine transform is performed in the unit of 8 × 8 pixels by the discrete cosine transformer 208, and the transform result is quantized by the quantizer 209 and output from the output 214. The quantized signal is inversely transformed by the inverse quantizer 210 and the inverse discrete cosine transformer 211 and stored in the motion compensation predictor 212. The intra-frame coding mode and the inter-frame coding mode have been described above. In H.261, it is possible to switch between intraframe and interframe modes for each macroblock. In this embodiment, the mode setter 205 is set to the interframe coding mode. 16x signal from memory A203
Read determination circuit B20 for each 16-pixel macro block
4, the dynamic range of each macroblock is obtained by the decision circuit B204 and compared with a threshold value. If the dynamic range is less than the threshold value, the motion compensation predictor 212
Is set to 0 and no motion vector detection is performed. When the dynamic range exceeds the threshold value, the motion compensation predictor 212 performs motion compensation detection after motion vector detection. The luminance value of each pixel in the macroblock is represented by 8-bit 256 gradations, and when the threshold value is set to a value of 50 or less, the macroblock which has little luminance change and is easy to perform pseudo matching in motion vector detection detects the motion vector. Without predicting the motion vector 0. Furthermore, in this method, macroblocks whose dynamic range is less than the threshold value do not detect motion vectors, so that the overall throughput increases. This is an extremely arithmetic operation in which the motion vector detection circuit obtains the sum of absolute differences between all pixels in the macroblock for each search in the search range ± 15 pixels, and obtains it in the entire search range to obtain the minimum distortion. This is because it is a large amount and takes much time as compared with other circuit parts, so that the overall throughput cannot be increased. However, in this method, the number of macroblocks for motion vector detection is reduced, so that the throughput is increased without changing other circuits.

(Embodiment 3) FIG. 3 is a block diagram showing an embodiment of a decision circuit for a videophone according to the present invention. The luminance value of each pixel of the macro block is input to the latch 301 every clock, and the output of the latch 301 is latch 30 every clock.
Enter in 2. The outputs of the latch 301 and the latch 302 are
Each bit is compared by the 3-bit EXNOR 303 from the MSB side, and when all 3 bits are equal, the output of the AND 304 becomes high. The luminance values of 16 × 16 pixels of the macroblock are sequentially input to the latch 301, and if the outputs of the AND 304 are all high, it is determined that the macroblock is below the threshold value. It is determined that it has been exceeded. If the number of bits to be compared is set to 2 bits, the possibility that the macro block will be judged to be below the threshold value increases,
If the number of bits to be compared is 4 bits or 5 bits, the number will be reduced. In this way, the comparison with the threshold value can be performed with an extremely simple circuit.

[0009]

As described above, according to the present invention, a macroblock which has a small dynamic range and is easy to be pseudo-matched in motion vector detection is not subjected to intra-frame coding or motion vector detection without motion vector detection.
Since the predictive coding is performed with, it is possible to prevent the generated code amount from increasing in the motion compensation prediction, and since there is a macro block that does not detect the motion vector, the overall throughput is increased, and the determination circuit is It has an effect that it can be realized with an extremely simple configuration.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a videophone of the present invention.

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

FIG. 3 is a block diagram showing an embodiment of a decision circuit of the videophone of the present invention.

[Explanation of symbols]

 Reference Signs List 101 camera 102 pre-processing circuit 103 memory A 104 determination circuit A 105 mode setter 106 switch 107 switch 108 discrete cosine transformer 109 quantizer 110 inverse quantizer 111 inverse discrete cosine transformer 112 motion compensation predictor 113 loop filter 114 Output 115 Information source encoder 201 Camera 202 Pre-processing circuit 203 Memory A 204 Judgment circuit B 205 Mode setter 206 Switch 207 Switch 208 Discrete cosine transformer 209 Quantizer 210 Inverse quantizer 211 Inverse discrete cosine transformer 212 Motion Compensation predictor 213 Loop filter 214 Output 215 Source encoder 301 Latch A 302 Latch B 303 EXNOR 304 AND

Claims (3)

[Claims] 1. A video input camera, a memory for recording one frame of a signal from the video input camera, and CCITT Recommendation H.264 for encoding a signal from the memory. 261
In a videophone having a compliant information source encoder, the signal of the memory is read for each macroblock of 16 × 16 pixels, the dynamic range of each macroblock is obtained and compared with a threshold value. If the dynamic range is less than or equal to the threshold, the information source encoder is set to intraframe coding mode, and if the dynamic range exceeds the threshold, the information source encoder is set to interframe coding mode. A videophone having a determination circuit A for setting. 2. A video input camera, a memory for recording one frame of a signal from the video input camera, and CCITT Recommendation H.264 for encoding the signal from the memory. 261
In a videophone having a compliant information source encoder, the signal of the memory is read for each macroblock of 16 × 16 pixels, the dynamic range of each macroblock is obtained and compared with a threshold value. If the dynamic range is less than or equal to the threshold, the source encoder is set to an interframe coding mode with a motion vector of 0, and if the dynamic range exceeds the threshold, the source encoder is activated. A videophone, comprising: a determination circuit B for setting an interframe coding mode with vector detection. 3. The determination circuit A for the videophone according to claim 1.
Alternatively, in the judging circuit B of the videophone according to claim 2,
MSB side N bits (N = 2 or 3 or 4 or 5) of luminance values of all pixels in a macro block of 16 × 16 pixels
Is the same value, it is determined that the dynamic range is less than or equal to the threshold value, and the MSB side N bits (N = 2 or 3 or 4 or 5) of the brightness value of all pixels in the macro block of 16 × 16 pixels are not the same value. In this case, the videophone has a determination circuit for determining that the dynamic range exceeds a threshold value.