JPH05103317A - Video signal transmitting device - Google Patents

Abstract

PURPOSE:To improve the quality of transmitted pictures by deciding quantizing sizes by blocks based on the difference between the local decoded value of transmitted picture data and original picture data. CONSTITUTION:Coefficient data S5 are found by orthogonally transforming video signals VD forming a unit block group GOB wit plural unit blocks MB and the data S5 are converted into quantized data S6 by quantization. Namely, local decoding means 23-25 reproduce the video signals VD by quantizing the data S5 based on the 1st quantization information S9 decided at every unit block group GOB and performing inverse quantization and inverse orthogonal transformation. A difference data detecting means 28 finds the difference between the decoded data S22 decoded by the means 23-25 and original video signals S1 corresponding to the data S22. A control means 22 sets the 2nd quantization information S24 at every unit block MM based on the detected results of the means 28. A quantizing means 8 sets the size STPS of quantization of the video signals VD based on the quantization information S9 and S24.

Description

Detailed Description of the Invention

[0001]

[Table of Contents] The present invention will be described in the following order. Field of Industrial Application Conventional Technology (FIGS. 5 to 8) Problems to be Solved by the Invention (FIGS. 5 to 8) Means for Solving the Problems (FIGS. 1 to 4) Action Example (FIGS. 1 to 1) Figure 4) Effect of the invention

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal transmission device, and a video signal for transmitting a high-quality video in a one-to-many transmission form such as in-broadcast transmission by an orthogonal transform method such as Discrete Cosine Transform. It is suitable for application to a transmission device.

[0003]

2. Description of the Related Art Conventionally, in a so-called video signal transmission system for transmitting a video signal composed of a moving image to a remote place such as a video conference system and a video telephone system, in order to efficiently use a transmission line, The video signal is coded by utilizing the correlation of (1) to improve the transmission efficiency of significant information.

For example, in the intra-frame coding process, as shown in FIG. 3, when each image PC1, PC2, PC3, ... Which constitutes a moving image is transmitted at time t = t1, t2, t3. The image data to be processed is one-dimensionally encoded and transmitted within the same scanning line. In addition, the inter-frame encoding process uses the autocorrelation of the video signal with respect to the time axis to sequentially adjoin the images PC1, PC2, and PC2.
, And the image data PC12, PC23 ... Which are the difference in pixel data between PC3.

As a result, the video signal transmission system high-efficiency-codes the images PC1, PC2, PC3, ... It can be sent to.

That is, as shown in FIG. 4, the image data transmission device 1 performs band limitation, transmission order conversion, etc. on the digitized input video signal VD by the preprocessing circuit 2 and outputs it as input image data S1. To do. Here, the image data sequentially sent as the input image data S1 is
It is extracted from the frame image data FRM by the method as shown in FIG.

One frame image data FRM is divided into, for example, 2 (horizontal direction) × 6 (vertical direction) block groups GOB as shown in FIG. 5A, and each block group GOB is shown in FIG. 11 (horizontal direction) x 3 (vertical direction) macroblock MBs as shown in (B)
, And each macroblock MB is shown in FIG.
As shown in (C), the luminance signal data Y1 for 8 × 8 pixels
Color difference signal data Cb and Cr which are color difference signal data corresponding to all pixel data of Y4.

At this time, the arrangement of the image data in the block group GOB is such that the image data is continuous in units of macro blocks MB, and in the macro block MB, the image data is continuous in units of minute blocks in the order of raster scanning. It is designed to do.

Here, the macroblock MB uses 16 × 16 pixel image data (Y1 to Y4) which are continuous in the horizontal and vertical scanning directions as one unit for the luminance signal, which corresponds to this. The two color difference signals are subjected to time-axis multiplexing processing after the data amount is reduced, and data of 16 × 16 pixels are assigned to each of the minute blocks Cr and Cb.

The difference data generation circuit 3 receives the input image data S
When the previous frame data S2 of the previous frame stored in the previous frame memory 4 is input together with 1, the difference between the previous frame data S2 and the input image data S1 is obtained to generate interframe encoded data (hereinafter referred to as interframe encoding mode). ), The difference data S3 is output via the switching circuit 5 to the discrete cosine transform (DCT) circuit 6 and the switching control circuit 7 together with the input image data S1.

The switching circuit 5 is controlled by the control signal S4 output from the switching control circuit 7, and when it is judged that there is a high possibility that a smaller amount of data can be transmitted by intra-field coding and transmission, The difference data S3 is output when it is determined that the input image data S1 is output as it is, and that the possibility of transmission with a small amount of data is high when the data is encoded and transmitted between frames.
The discreet cosine transform circuit 6 performs discrete cosine transform on the input image data S1 or the difference data S3 in a unit of a small block in order to utilize the two-dimensional correlation of the video signal, and the resulting coefficient data S5 is quantized by a quantizing circuit 8
It is designed to output to.

The quantization circuit 8 is a block group GOB.
The converted data S5 is quantized with a quantization step size determined for each, and the quantized data S6 obtained at the output end is quantized by a variable length coding circuit (VLC) 9
And the inverse quantization circuit 12. Here, the variable length coding circuit 9 performs variable length coding processing on the quantized data S6 and supplies it to the transmission buffer memory BM10 as transmission data S7.

The transmission buffer memory 10 stores the transmission data S
7 is once stored in the memory and then output to the transmission line 11 as output data S8 at a predetermined timing, and the quantization control signal S9 in block group GOB units is quantized according to the amount of residual data remaining in the memory. The circuit 8 is fed back to control the quantization step size. As a result, the transmission buffer memory 10
The amount of data generated as the output data S8 is adjusted to maintain an appropriate amount of data (data amount that does not cause overflow or underflow) in the memory.

When the remaining amount of data in the transmission buffer memory 10 is increased to the allowable upper limit, the transmission buffer memory 10
Reduces the data amount of the quantized data S6 by increasing the step size of the quantized step size STPS (FIG. 6) of the quantization circuit 8 in accordance with the quantized control signal S9. On the contrary, when the remaining amount of data in the transmission buffer memory 10 is reduced to the allowable lower limit value, the transmission buffer memory 10 reduces the step size of the quantization step size STPS of the quantization circuit 8 by the quantization control signal S9. This increases the data amount of the quantized data S6.

The inverse quantization circuit 12 inversely quantizes the quantized data S6 sent from the quantization circuit 8 into a representative value and converts the quantized data S6 into inverse quantized data S10, and outputs the output data S8 before conversion in the quantization circuit 8. Of the inverse quantized data S10, the inverse quantized data S10
The signal is supplied to an inverse discrete cosine transform (T) circuit 13. The inverse cosine inverse transform circuit 13 transforms the inverse quantized data S10 decoded by the inverse quantizer circuit 12 into the decoded image data S11 by the inverse transform process of the discrete cosine transform circuit 6, and the previous frame data generation circuit. 14 and switching circuit 15.

As a result, the discrete cosine inverse conversion circuit 13 outputs the input image data S1 or the difference before the conversion of the output data S8 output through the transmission line 11 and reproduced on the receiving side in the discrete cosine conversion circuit 6. The data S3 can be decoded on the transmission side. That is, the discrete cosine inverse conversion circuit 13
Reproduces the input image data S1 when the video signal VD is intra-field encoded and transmitted, while the difference data S3 is reproduced when the video signal VD is inter-frame encoded and transmitted. Is designed to be restored.

The previous frame data generating circuit 14 adds the previous frame data S2 fed back from the previous frame memory 4 and the decoded image data S11 to output data S
Restore the image data of the previous frame output as 8,
By outputting to the previous frame memory 4 via the switching circuit 15, the image transmitted to the receiving side is sequentially reproduced and stored in the previous frame memory 4. Here, the switching circuit 15 is switched and controlled by the control signal S4 delayed by the time required from the discrete cosine conversion of the video signal VD to the inverse discrete cosine conversion through the delay circuit 16. ing.

[0018]

However, in the conventional image data transmission apparatus 1, the amount of generated information data for each block group GOB generated in the quantization circuit 8 is averaged based on the remaining amount of data in the transmission buffer memory 10. Therefore, when a pattern that is likely to be distorted and a pattern that is not likely to be distorted are mixed in the block group GOB, deterioration of the image quality is easily visible in the portion of the pattern that is likely to be distorted, and the image quality is not constant. There was a problem.

In the case of a picture in which the amount of transmitted information increases and decreases locally and abruptly, such as an image of a rotating water turbine, the amount of image information is flat in one block among a plurality of blocks MB forming the block group GOB. Since the average part and the detailed part are included, if the quantization step size is set on average, the distortion will be locally concentrated on the block containing the water wheel's splashes and the splashes will appear blurred. Or, there is a possibility that a block-shaped distortion may be visually observed in the flat part around the block. The distortion cosine conversion method is characterized in that the distortion is diffused throughout the block, and the way the distortion is generated varies depending on the image of the image to be transmitted, and the distortion tends to be non-uniform. In a transmission device, it is important that the image quality be uniform regardless of the nature of the transmitted image.

Further, since the visual characteristics of the luminance signal are not linear, even if the amount of distortion is the same, the distortion may be easily perceived or difficult to perceive due to the level of the luminance level. The distortion caused by the visibility is particularly important in a high image quality transmission apparatus in which it is desirable that the image quality is uniform regardless of the property of the transmitted image.

The present invention has been made in consideration of the above points, and even in the case where an input image having a large change in which an image easily distorted and an image hardly distorted are mixed is input,
By adjusting the amount of generated information directly corresponding to the input image, the image quality of the entire image can be further improved and transmitted.

[0022]

In order to solve such a problem, in the first invention, a video signal VD forming a unit block group GOB is orthogonally transformed by a plurality of unit blocks MB to obtain coefficient data S5, In the video signal transmission device 20 which quantizes the coefficient data S5 and converts it into the quantized data S6, the first is determined for each unit block group GOB.
The coefficient data S5 is quantized and inversely quantized based on the quantization information S9 of the above-mentioned video signal V.
The local decoding means 23, 24, 25 for restoring D, and the differential data detection means 28 for obtaining the difference between the decoded data S22 decoded by the decoding means 23, 24, 25 and the original video signal S1 corresponding to the decoded data S22. Based on the detection result of the difference data detection means 28, the control means 22 for setting the second quantization information S24 for each unit block MB, the first quantization information S9 and the second quantization information S24 are set. Based on the above, the quantization means 8 for setting the quantization size STPS of the video signal VD is provided.

In the second invention, the control means 22
Is a distortion evaluation unit 3 for obtaining a distortion evaluation amount for each unit block MB based on the detection result of the difference data detection unit 28.
0, 31 and the reference distortion evaluation means 33 for obtaining the reference distortion evaluation amount S33 for each unit block group GOB based on the first quantization information S9, and output from the distortion evaluation means 30, 31 and the reference distortion evaluation means 33. Based on the comparison result S34 of the distortion amount comparison means 32 and the distortion amount comparison means 32 which compares the distortion evaluation amount S32 and the reference distortion evaluation amount S33, the unit block MB
And the quantization information control means 34 for controlling the second quantization information S24 for each of them.

Further, in the third invention, the control means 4
1 is distortion evaluation means 42, 31 for obtaining a distortion evaluation amount S32 for each unit block MB based on the detection result of the differential data detection means 28 and the level S41 of the original video signal corresponding to the locally decoded data S22; Based on the quantized information S9 of the reference block evaluation amount S3 for each unit block group GOB.
Reference strain evaluation means 33 for determining 3 and strain evaluation means 42, 3
1 and the strain evaluation amount S3 output from the reference strain evaluation means 33.
2 and the strain amount comparing means 32 for comparing the reference strain evaluation amount S33.
And a quantized information control means 34 for controlling the second quantized information S24 for each unit block MB based on the comparison result of the distortion amount comparison means 32.

[0025]

In the same unit block group GOB, a video signal VD that easily causes image quality deterioration and a video signal V that hardly causes image quality deterioration
When D is mixed, coefficient data obtained by orthogonally transforming the video signal VD is locally decoded to obtain a difference S from the original video signal S1.
23, and corresponds to the corresponding coefficient data based on the second quantization information S24 determined for each unit block MB set by the difference S23 and the first quantization information S9 determined for each unit block group GOB. By controlling the quantization characteristic, it is possible to send the transmission data without deteriorating the image quality even when an image in which the distortion is locally and rapidly increased is input.

[0026]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail with reference to the drawings.

In FIG. 1 in which parts corresponding to those in FIG. 4 are designated by the same reference numerals, numeral 20 indicates an image data transmission apparatus as a whole, which includes an inverse quantization circuit 12, a discrete cosine inverse conversion circuit 13, and previous frame data generation. First consisting of circuit 14
In addition to the local decoding circuit system 21A, the second local decoding circuit system 21B is provided, and each block of the transmission image is based on the output result output from the first and second local decoding circuit systems 21A and 21B. It has the same configuration except that it has a quantization parameter control circuit 22 for controlling the quantization parameter of.

The second local decoding circuit system 21B obtains the decoded value of the coefficient data S5 output from the discrete cosine conversion circuit 6 before the quantization circuit 8 quantizes and outputs the coefficient data S5. The distortion of the original image data is detected.

Here, the second local decoding circuit system 21B inputs the coefficient data S5 output from the discrete cosine transform circuit 6 to a quantizing / inverse quantizing circuit (QQ-) 23 which is a read-only memory or the like. The quantizing / inverse quantizing circuit (QQ-) 23 quantizes the coefficient after the discrete cosine transform based on the quantizing characteristic determined by the quantizing control signal S9 for each block group GOB fed back from the transmission buffer memory 10. (That is, classification) and inverse quantization (that is, representative value), and output to the discrete cosine inverse conversion circuit 24.

Discrete cosine inverse conversion circuit 24
Outputs the inverse conversion data S21 obtained by inversely converting the representative value to the adder circuit 25 and the switching circuit 26.
Here, the adder circuit 25 adds the inversely transformed data S21 to the locally decoded data S2 of the previous frame stored in the previous frame memory 4 so that the data is transmitted at the quantization step size determined by the quantization control signal S9. The image data is locally decoded and output to the switching circuit 26.

The switching circuit 26 has a control signal S4 delayed by the time required for the second local decoding circuit system 21B to process the image data from the discrete cosine conversion circuit 6 to the discrete cosine inverse conversion circuit 24. Is input via the delay circuit 27, and is output to the distortion amount calculation circuit 28 depending on whether the current image data transmitted via the transmission path 11 is inter-field encoded processing data or inter-frame encoded processing data. The locally decoded data S22 to be converted is switched.

The distortion amount calculation circuit 28 calculates the difference between the input image data S1 as an original image input via the delay circuit 29 and the locally decoded data S22 input via the switching circuit 26, and thereby the same. The distortion amount for the sample is calculated, and the distortion amount is output to the quantization parameter control circuit 22 as distortion data S23. Here, the delay circuit 29 is
FIFO (first in first out) memory configuration,
The output of the control signal S4 is delayed by the time required for the local decoding circuit 25 to process the input image data S1 via the discrete cosine conversion circuit 6.

The quantization parameter control circuit 22 receives the distortion data S23 input from the distortion amount calculation circuit 28 and the block group GO input from the transmission buffer memory 10.
Based on the quantization control signal S9 for each B, the quantization control signal S24 for controlling the quantization step size for each block of the transmitted image data is output to the quantization circuit 8. As a result, the quantization parameter control circuit 22 can locally control the quantization step size for each block prior to transmission within a range that does not affect the smoothing of the generated information amount of image data. As a result, a block having a large distortion within the same block group GOB can have a fine quantization step size, while a block having a small distortion can have a large quantization step size, so that the quality of a transmitted image can be kept constant. Has been done.

That is, the quantization parameter control circuit 22
2, the distortion data S23 input from the distortion amount calculation circuit 28 is input to the absolute value circuit 30, the absolute value of the distortion with respect to the original image data is obtained, and this is output as the absolute difference data S31. It is done like this. The integrator circuit 31 obtains the sum of the absolute difference data S31 input from the absolute value circuit 30 for each block, and supplies it to the subtractor circuit 32 as the distortion sum total data S32.

The subtraction circuit 32 calculates the difference from the standard distortion data S33 for each block input from the standard distortion generation circuit 33 composed of a ROM or the like, and the difference data S34.
Is output to the quantization parameter setting circuit 34. Here, the standard distortion generating circuit 33, based on the control parameter for each block group GOB input from the transmission buffer memory 10 as the quantized control signal S9,
The sum of absolute values of standard or average strain that will occur for each block is estimated in advance, and the estimated value is used as standard strain data S3.
3 is output.

Here, the quantization parameter setting circuit 34
The quantization circuit 8 and the inverse quantization circuit 1 are set as the control data S24 for each block, which is a control parameter for determining the quantization step size of each block so that the difference data S34 between the total distortion data S32 and the standard distortion data S33 becomes small.
2 to output the second local decoding circuit system 21B.
When the image data corresponding to the block group GOB processed in (1) is output via the transmission path 11, the image quality is controlled to be substantially constant. Also, the first local decoding circuit system 21A
The inverse quantization circuit 12 of FIG.
The quantized data S6 is inversely quantized on the basis of 4.

Incidentally, the image data transmission device 20 uses the conversion data S output from the discrete cosine conversion circuit 6.
5 is delayed by a delay circuit 36 formed of a FIFO memory for a predetermined time, and then supplied to the quantization circuit 8. Here, the delay circuit 36 determines the time required for the second local decoding circuit system 21B and the quantization parameter control circuit 22 to process the image data, that is, the quantization / inverse quantization circuit 23 to the quantization parameter setting circuit 34. The converted data S5 is delayed and supplied to the quantization circuit 8 for the time required to complete the signal processing, whereby the image data generation circuit 20 is based on the nature of the image actually transmitted. Feedforward processing is possible.

The image data generating circuit 20 also supplies the previous frame data S supplied from the previous frame memory 4 to the previous frame data generating circuit 14 of the first local decoding circuit system 21A.
2 is input through a delay circuit 37 which is a FIFO memory. Here, the delay circuit 37 delays and outputs the previous frame data by the time required for the processing time of the second local decoding circuit system 21B and the quantization parameter control circuit 22. The previous frame data actually output via the above can be decoded.

In the above structure, a block corresponding to a boundary portion where a background portion where deterioration of image quality is unlikely to occur and a splashing portion of water turbine where deterioration of image quality is likely to occur coexist, like an image of a water turbine against the sky. When the video signal VD of the group GOB is input to the preprocessing circuit 2, the preprocessing circuit 2 converts the video signal VD into 8-bit input image data S1 and supplies it to the difference data creating circuit 3 and the delay circuit 29. .. The difference data creation circuit 3 obtains interframe difference data S3 between the current input image data S1 input from the preprocessing circuit 2 and the corresponding block group GOB of the previous frame supplied from the previous frame memory 4, and performs the discrete cosine conversion. In the circuit 6, the two-dimensional discrete cosine transform is performed for each block.

Here, the discrete cosine conversion circuit 6
Is the time required for the second local decoding circuit system 21B and the quantization parameter control circuit 22 to process the converted data S5 by way of the delay circuit 36.
Is delayed and supplied to the quantization circuit 8. As described above, while the delay circuit 36 delays the conversion data S5 and delays the quantization of the conversion data, the second local decoding circuit system 21B and the quantization parameter control circuit 22 are in the block group currently being transmitted. Each block, which is a constituent unit of GOB, is a block corresponding to the background portion in which image quality deterioration is unlikely to occur, or a block corresponding to the boundary portion between the background and the splash of the water turbine is likely to occur. Determine the quantization step size for each block.

That is, the second local decoding circuit system 21B quantizes the converted data S5 via the quantizing / inverse quantizing circuit 23 by the quantizing control signal S9 which determines the quantizing step size of the block group GOB, and quantizes it. After the subsequent transformed data S5 is inversely quantized again, the inversely quantized representative value is inversely transformed by the discrete cosine inverse transform circuit 24. At this time, when the second local decoding circuit system 21B obtains the local decoding value based on the quantization accuracy for each block group GOB of the current frame from the local decoding value of the previous frame in the local decoding circuit 25, the distortion amount is output via the switching circuit 26. Calculation circuit 28
Image data S22 of the current block group supplied to and decoded by
And the input image data S1 that is the original image are obtained, and the distortion data S23 is output to the quantization parameter control circuit 22.

At this time, the quantization parameter control circuit 22
When the absolute value circuit 30 obtains the absolute value of the distortion data S23 for each block, the integrating circuit 31 calculates the total sum of the distortions for each block, and the demodulated value of each block is the original value for the current quantization step size. Calculate how much the image data will actually deviate. At the same time, the quantization parameter control circuit 22 causes the transmission buffer memory 10
Block group G that determines the quantization size so that the amount of data stored in the stream does not overflow or underflow
When the control parameter for each OB is input as the quantized control signal S9, the amount of distortion that will occur for each block is estimated,
The standard distortion generating circuit 33 outputs the subtraction circuit 32.

After this, the quantization parameter control circuit 22
When the subtraction circuit 32 finds the difference between the actual distortion amount and the standard distortion data S33, the difference data S34 is supplied to the quantization parameter setting circuit 34. Here, the quantization parameter setting circuit 34 locally increases the distortion amount of each block with respect to the distortion amount predicted by the average quantization step size of the block group GOB from the increase / decrease of the difference data S34. It is determined whether or not there is.

For example, in the block group GOB having many blocks corresponding to the background area image, the quantization step size is set large in the transmission buffer memory 10 because the amount of distortion is small, but the block corresponding to the boundary portion between the water wheel and the background area is set. Then, a larger amount of distortion occurs than the amount of distortion estimated by the current quantization step size. In such cases,
The quantization parameter setting circuit 34 reduces the quantization step size of the corresponding block to reduce the distortion amount within the range where the transmission buffer memory 10 does not overflow or underflow, and reduces the image quality at the boundary between the water wheel and the background. The quantization control signal S24 is output to the quantization circuit 8 and the inverse quantization circuit 12 so that the deterioration does not occur, and the control parameter for each block is set to a predetermined value.

Thereafter, the quantizing circuit 8 calculates, based on the quantization parameter for each block calculated prior to the image data of the block group GOB delayed by the delay circuit 36 and the quantization parameter for each block group GOB, The quantization step size for each block constituting the block group GOB is controlled and supplied to the variable length coding circuit 9, and after the variable length coding processing, it is output to the transmission line 11 via the transmission buffer memory 10. At this time, the dequantization circuit 12 of the first local decoding circuit system 21A dequantizes the quantized data S6 with the actual quantization step size set by the quantization control signals S9 and S24. 4, the actually transmitted image data is decoded and stored on the transmission side, and the same operation is repeated thereafter.

According to the above construction, prior to the quantization of the transmitted image data, the locally decoded value of the actually transmitted image data is obtained in advance, and the distortion amount of this locally decoded value with respect to the original image data is calculated. After obtaining for each block, by locally controlling the quantization step size for each block based on the distortion amount, an image in which the distortion amount easily and locally increases and decreases within the same block group is input. Even in such a case, it is possible to more easily balance the generated distortion information amount of the entire block group GOB and prevent the local deterioration of the image quality.

In this case, the same block group GOB is also used.
Since the total amount of strain generated in each block can be close to the standard strain estimated in advance, it is possible to effectively avoid the possibility that the strain distribution becomes non-uniform within the block group GOB.

In the above embodiment, the case where the quantization parameter control circuit 22 obtains the absolute value of the difference data S23 for the same sample of the input image signal S1 as the original image data and the locally decoded image data S22 has been described. However, the present invention is not limited to this, and the sum of squares of each difference data S23 may be obtained, and the present invention can be applied to various cases such as non-linearly weighted values.

In the above embodiment, the quantization parameter control circuit 22 calculates the sum of absolute values of the standard distortion amounts for each block based on the quantization control signal S9 indicating the quantization control parameter for each block group GOB. Although the case of estimation is described, the present invention is not limited to this, and a standard sum of squares of strain amounts or the like may be estimated.

Further, in the above embodiment, the case where the processing circuit having the configuration shown in FIG. 2 is used as the quantization parameter control circuit 22 has been described, but the present invention is not limited to this, and the input image signal composed of the original image data is used. Difference data S2 for the same sample of S1 and locally decoded image data S22
2 and control parameter S9 for each block group GOB
The present invention can be widely applied to various processing circuits that control the quantization parameter in block units based on the quantization control signal.

Further, in the above-mentioned embodiment, the case where the image of the water turbine is signal-processed and transmitted with the sky as the background has been described, but the present invention is not limited to this, and is suitable for transmitting an input image that changes drastically. Is.

Further, in the above embodiment, the difference data S2 between the input video signal S1 and the locally decoded image data S22 is obtained.
2 and the case where the quantization parameter in block units is controlled based on the quantization control signal S9 which is a control parameter for each block group GOB, the present invention is not limited to this, and the filter characteristics and the like may be controlled. May be.

Further, in the above embodiment, the difference data S2 between the input video signal S1 and the locally decoded image data S22 is obtained.
2 and the case where the quantization parameter in block units is controlled based on the quantization control signal S9 that is the control parameter for each block group GOB, the present invention is not limited to this, and the transmission image data is processed by signal processing. It can also be widely applied when other processed signals are used.

Further, in the above embodiment, the case where the image data is weighted using the image transmission device 20 and the quantization parameter control circuit 22 shown in FIGS. 1 and 2 has been described, but the present invention is not limited to this. It is also possible to use the image transmission device 40 and the quantization parameter control circuit 41 shown in FIGS. 3 and 4 in which parts corresponding to those in FIGS.

Here, the quantization parameter control circuit 41 is
The distortion data S23 is input from the distortion amount calculation circuit 28, and the luminance signal S41 of the original signal is input from the delay circuit 29. The quantization parameter control circuit 41 has a weighting circuit 42 including a ROM,
Distortion data S23 is weighted based on the luminance signal S41,
The absolute value of the strain data after weighting is weighted to the difference data S
The data is output to the integration circuit 31 as 42.

At this time, the weighting circuit 42 non-linearly weights the distortion data S23 so as to match the human visual characteristics. In the case of this embodiment, the standard distortion generating circuit 33 will be generated in each block at the average luminance signal level based on the control parameter for each block group GOB inputted as the quantized control signal S9 from the transmission buffer memory 10. A standard or average sum of absolute values of distortion is estimated and output as standard distortion data S33.

At this time, the quantization parameter setting circuit 3
When the total distortion data S32 is larger than the standard distortion data S33, 4 outputs the control data S24 so as to improve the quantization accuracy, and when the total distortion data S32 is smaller than the standard distortion data S33, the quantization is performed. The control data S24 is output so as to deteriorate the accuracy.
As a result, the image transmission device 40 can achieve uniform distortion including human visual perception characteristics, and can further effectively utilize the information amount and achieve high image quality transmission.

Further, although the weighting circuit 42 weights the distortion data S23 in a non-linear manner in the above-mentioned embodiment, the present invention may be weighted linearly instead.

Further, in the above-mentioned embodiment, the case where the luminance signal is inputted for all the bits has been described, but the present invention is not limited to this, and the signal of 3 to 5 bits on the most significant bit side is inputted. Is also good.

[0060]

As described above, according to the present invention, the image data to be transmitted is decoded based on the first quantization information determined for each block group, and the difference between the decoded data and the original image data is decoded. It controls the second quantization information determined for each block according to the data, and controls the quantization size of the image data actually transmitted based on the first quantization information and the second quantization information. As a result, even when image data in which the distortion amount is likely to locally increase or decrease is input in the same block group, the image data can be transmitted without degrading the image quality.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an image data transmission device according to the present invention.

FIG. 2 is a block diagram for explaining the quantization parameter control circuit.

FIG. 3 is a block diagram showing an image data transmission device according to another embodiment.

FIG. 4 is a block diagram for explaining the quantization parameter control circuit.

FIG. 5 is a schematic diagram for explaining intraframe / interframe encoding processing.

FIG. 6 is a block diagram for explaining a conventional image data transmission device.

FIG. 7 is a schematic diagram showing the structure of frame image data.

FIG. 8 is a schematic diagram for explaining a quantization step.

[Explanation of symbols]

20, 40 ... Image data transmission device, 21A, 21B ...
... Local decoding circuit system, 22, 41 ... Quantization parameter control circuit, 23 ... Quantization / inverse quantization circuit, 24 ... Discrete cosine inverse conversion circuit, 25 ... Local decoding circuit, 28 ... Distortion amount Calculation circuit, 33 ... Standard distortion generation circuit,
34 ... Quantization parameter setting circuit.

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] August 17, 1992

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] 0002

[Correction method] Change

[Correction content]

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal transmission device, and more particularly to a video signal transmission device for transmitting a high quality image in a one-to-many transmission mode such as broadcasting by orthogonal transformation such as Discrete Cosine Transform. It is suitable for application.

[Procedure Amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0004

[Correction method] Change

[Correction content]

For example, in the intra-frame coding process, as shown in FIG. 5, when transmitting each image PC1, PC2, PC3, ... Which constitutes a moving image at time t = t1, t2, t3. The image data to be processed is one-dimensionally encoded and transmitted within the same scanning line. In addition, the inter-frame encoding process uses the autocorrelation of the video signal with respect to the time axis to sequentially adjoin the images PC1, PC2, and PC2.
, And the image data PC12, PC23 ... Which are the difference in pixel data between PC3.

[Procedure 3]

[Document name to be amended] Statement

[Correction target item name] 0006

[Correction method] Change

[Correction content]

That is, as shown in FIG. 6, the image data transmission apparatus 1 performs band limitation, transmission order conversion, etc. on the digitized input video signal VD by the preprocessing circuit 2 and outputs it as input image data S1. To do. Here, the image data sequentially sent as the input image data S1 is
It is extracted from the frame image data FRM by the method as shown in FIG.

[Procedure amendment 4]

[Document name to be amended] Statement

[Correction target item name] 0007

[Correction method] Change

[Correction content]

One frame image data FRM is divided into, for example, 2 (horizontal direction) × 6 (vertical direction) block groups GOB as shown in FIG. 7A, and each block group GOB is shown in FIG. 11 (horizontal direction) x 3 (vertical direction) macroblock MBs as shown in (B)
, And each macroblock MB is shown in FIG.
As shown in (C), luminance signal data Y1 for 8 × 8 pixels
Color difference signal data Cb and Cr, which are color difference signal data corresponding to all pixel data of Y4.

[Procedure Amendment 5]

[Document name to be amended] Statement

[Correction target item name] 0014

[Correction method] Change

[Correction content]

When the remaining amount of data in the transmission buffer memory 10 is increased to the allowable upper limit, the transmission buffer memory 10
Reduces the data amount of the quantized data S6 by increasing the step size of the quantized step size STPS (FIG. 8) of the quantization circuit 8 in response to the quantized control signal S9. On the contrary, when the remaining amount of data in the transmission buffer memory 10 is reduced to the allowable lower limit value, the transmission buffer memory 10 reduces the step size of the quantization step size STPS of the quantization circuit 8 by the quantization control signal S9. This increases the data amount of the quantized data S6.

[Procedure correction 6]

[Document name to be amended] Statement

[Name of item to be corrected] 0019

[Correction method] Change

[Correction content]

In the case of a picture in which the amount of transmitted information increases and decreases locally and abruptly, such as an image of a rotating water turbine, the amount of image information is flat in one block among a plurality of blocks MB forming the block group GOB. Since the average part and the detailed part are included, if the quantization step size is set on average, the distortion will be locally concentrated on the block containing the water wheel's splashes and the splashes will appear blurred. Or, there is a possibility that a block-shaped distortion may be visually observed in the flat part around the block. The distortion cosine conversion method is characterized in that the distortion is diffused throughout the block, and the way the distortion is generated varies depending on the image of the image to be transmitted, and the distortion tends to be non-uniform. In a transmission device, it is important that the image quality be uniform regardless of the nature of the transmitted image, so this distortion becomes a serious problem.

[Procedure Amendment 7]

[Document name to be amended] Statement

[Correction target item name] 0020

[Correction method] Change

[Correction content]

Further, since the visual characteristics of the luminance signal are not linear, even if the amount of distortion is the same, the distortion may be easily perceived or difficult to perceive due to the level of the luminance level. The distortion caused by the visibility is a particularly important problem in a high image quality transmission apparatus in which it is desirable that the image quality is uniform regardless of the property of the transmitted image.

[Procedure Amendment 8]

[Document name to be amended] Statement

[Name of item to be corrected] 0029

[Correction method] Change

[Correction content]

The second local decoding circuit system 21B in this case, the discrete cosine transform circuit quantization / inverse quantization circuit 6 the coefficient data S5 outputted from the consisting of a read only memory or the like - to enter the 23 ^(QQ). Quantization / inverse quantization circuit (QQ ^-) 23 based on the quantization characteristic which is determined by the blow stick group GOB every quantization control signal S9 which is fed back from the transmission buffer memory 10, quantizes the coefficient after discrete cosine transform (That is, classification) and inverse quantization (that is, representative value), and output to the discrete cosine inverse conversion circuit 24.

[Procedure Amendment 9]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 6

[Correction method] Change

[Correction content]

[Figure 6] ─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] November 16, 1992

[Procedure Amendment 9]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 6

[Correction method] Change

[Correction content]

[Figure 6]

Claims (3)

[Claims] 1. A video signal transmission apparatus for orthogonally transforming a video signal forming a unit block group with a plurality of unit blocks into coefficient data, quantizing the coefficient data, and converting the coefficient data into quantized data. Based on the first quantization information determined for each block group, the coefficient data is quantized and inversely quantized, and the local decoding means performs inverse orthogonal transformation, and the locally decoded data decoded by the local decoding means. And difference data detecting means for obtaining a difference between the original video signal corresponding to the local decoded data, and control means for setting the second quantization information for each unit block based on the detection result of the difference data detecting means. And a quantization means for setting a quantization size of the video signal based on the first quantization information and the second quantization information. Video signal transmission device. 2. The control means, based on the detection result of the difference data detection means, calculates the distortion evaluation amount for each unit block, and the unit based on the first quantization information. Reference strain evaluation means for obtaining a reference strain evaluation amount for each block group, a strain amount comparison means for comparing the strain evaluation amount and the reference strain evaluation amount output from the strain evaluation means and the reference strain evaluation means, and The video signal transmission apparatus according to claim 1, further comprising: quantization information control means for controlling the second quantization information for each unit block based on the comparison result of the distortion amount comparison means. 3. The distortion evaluating means for obtaining the distortion evaluation amount for each unit block based on the detection result of the difference data detecting means and the level of the original video signal corresponding to the locally decoded data. Reference distortion evaluation means for obtaining a reference distortion evaluation amount for each of the unit block groups based on the first quantization information, and the distortion evaluation amount and the reference output from the distortion evaluation means and the reference distortion evaluation means. A distortion amount comparing means for comparing the distortion evaluation amounts, and a quantization information control means for controlling the second quantization information for each unit block based on the comparison result of the distortion amount comparing means. The video signal transmission device according to claim 1.