JPH03124184A - Cell abort compensation picture decoding system - Google Patents

Abstract

PURPOSE:To output a picture data with less error in the coefficient of a low frequency by compensating and outputting a coefficient in one preceding frame when cell abort exists when a difference of the information representing the relation of quantity of each coefficient transmitted from a transmission side is large. CONSTITUTION:An abort compensation means 13 provided in the inside of a decoding section 17 uses information representing the relation of quantity of a difference of a coefficient sent from a coefficient quantity relation transmission means 11 at a transmission side as to a block corresponding to an aborted cell when cell abort exists so as to replace a coefficient with a small difference among coefficients in a group of minimum frequency into a coefficient corresponding to in a block of one preceding frame to the frame including the block and so as to replace a coefficient corresponding to the block as to a larger difference into a coefficient transmitted from the coefficient quantity relation transmission means 11 and outputs the result to an inverse orthogonal conversion section 12.

Description

[Detailed Description of the Invention] [Summary] In an image data transmission system that orthogonally transforms input image data, encodes coefficients in the orthogonal transform coefficient domain, and transmits them in cell units, when a cell is discarded, the coefficient domain Regarding the cell discard compensation image decoding method that compensates for multiple coefficients that represent low frequency components among the coefficients, the transmitting side compares the magnitude relationship of the corresponding low frequency coefficients of the blocks of the current frame and the previous frame, and detects when the difference is large. Sometimes, by sending compensation information to the receiving side, when cell discard occurs, the input image data is orthogonally transformed by an orthogonal transform unit with the aim of compensating for coefficients with a large difference when cell discard occurs. In an image data transmission system having an inverse orthogonal transform unit that encodes and transmits coefficients in the orthogonal transform coefficient domain in cell units and performs inverse orthogonal transform on the receiving side, the coefficients in the orthogonal transform coefficient domain are A plurality of coefficients representing frequency components are grouped into a plurality of groups corresponding to a plurality of frequency bands, and a transmission side of the image data transmission system is provided with a coefficient in the group of the lowest frequency among the plurality of groups in the block currently being transmitted. and a plurality of coefficients corresponding to the plurality of coefficients in a block corresponding to the block in a frame temporally one frame before the frame including the block, respectively. coefficient magnitude relationship transmitting means for transmitting information indicating whether the difference is large or small and the value of the coefficient in the current block for the coefficient with the large difference to the receiving side of the previous image data transmission system; On the receiving side of the system, the presence or absence of cell discard is detected, and when there is sale discard, the coefficient magnitude relationship between the block corresponding to the discarded cell and the coefficient magnitude relationship between the coefficients transmitted from the transmitting means For coefficients in the lowest frequency group that have a small difference, the corresponding coefficient data in the block corresponding to the block that is one frame before the frame containing the block, and also the coefficient data that has a large difference. With respect to the coefficients, the block is configured to include a discard compensation means for outputting data in which the corresponding coefficient in the block is replaced with the coefficient value transmitted from the coefficient magnitude relation transmitting means to the inverse orthogonal transform section.

[Industrial Application Field] The present invention relates to an image data transmission system, and more specifically, the present invention relates to an image data transmission system in which input image data is orthogonally transformed, coefficients in an orthogonal transform coefficient domain are encoded, and transmitted in units of cells. The present invention relates to a cell discard compensation image decoding method that compensates for a plurality of coefficients representing low frequency components among coefficient domain coefficients when discard occurs.

[Conventional technology]

Recently, the digitization of image equipment and the transmission method of image digital signals have been widely developed and researched in many application fields such as videophones. Generally, image data has a very large amount of information. For example, if current televisions were to be digitized as is, a transmission capacity of 100 Mb/s would be required, and image data has 1,500 times more information than audio. On a television, the screen changes 30 times per second, and frames change every 1/30 of a second. Pixels are generally arranged two-dimensionally on one screen corresponding to one frame,
Even if the level of one pixel is represented by 8 bits, for example, the amount of image data for one frame is extremely large.

However, even when frames are switched every 1/30 seconds on a television, the content of the pictures in each frame often does not change significantly. In other words, there are many parts of the image that are completely static, such as the sky and background, and there are few parts that change from frame to frame. Therefore, one method for compressing and encoding image data is an interframe encoding method.

In the interframe coding method, a difference in level data is determined for each corresponding pixel between two temporally consecutive frames, and the difference is subjected to, for example, Huffman encoding and sent onto a transmission path. If the level for most pixels does not change, the difference for those pixels will be 0, and the difference can be represented by one bit. Pixels with large level differences require codes of 8 bits or more, but by coding only these differences, the amount of information to be transmitted can be greatly compressed.

On the other hand, a method of encoding data within one frame as is is called an intraframe encoding method.

Regardless of whether interframe coding or intraframe coding is used, two operations, sampling and quantization, are basically required for digital representation of an image signal.There are two methods for sampling zero images. The first is to represent an image by level values for discrete sequences of dots corresponding to pixels within a frame.

The second method is to perform orthonormal expansion of a level value function (image function) defined on the XY plane corresponding to the frame surface, and use the expansion coefficients as sample values.

FIG. 9 is an overall block diagram of an image transmission system using orthonormal transformation, which is the second method. In FIG. 9, the input image is divided into one block, orthogonally transformed in 2, and then 3
The signal is quantized at 4, Huffman coded at 4, and sent out from the sending side to the transmission path 5. Then, on the receiving side, the signal from the transmission path 5 is decoded at 6, subjected to inverse orthogonal transformation at 7, converted into inverse blocks at 8, and output as image data. Here 1
Blocking is not done by orthogonally transforming two-dimensional KXK pixel data (for one screen), but by 4×4
This is done in order to shorten the conversion time by determining the conversion coefficients for each block, with approximately 16×16 pixels as one block. Examples of the types of orthogonal transforms mentioned in 2 include Hadamard transform, cosine transform, and Karhunen-Loeve transform, but in recent years, the discrete cosine transform (DCT), which will be described later, is a discrete representation of cosine transform.
is often used.

The orthogonal transform encoding method used in FIG. 9 focuses on the correlation between small areas within the screen, and orthogonal transform is performed using the pixels of the small area as a numerical string. The resulting conversion coefficients correspond to frequency components and represent each component from low frequency to high frequency. In image signals, low-frequency components are generally large in power and high-frequency components are small in power. Therefore, by performing quantization, which assigns a large number of bits to low-frequency components and a small number of bits to high-frequency components, it is possible to reduce transmission. The amount of information required is reduced.

FIG. 10 is a diagram showing a transform coefficient area after performing orthogonal transform on image data in an l block composed of 4×4 pixels. In this conversion coefficient region, the upper leftmost region 9 at -a represents the DC component of the signal, and the coefficients in the other regions all represent the AC components. The AC components correspond to each frequency from low frequency to high frequency, and the frequency becomes higher toward the lower right.

[Problem to be solved by the invention]

In the image data transmission system using orthogonal transformation as described above, regardless of whether intra-frame or inter-frame encoding is used, image data is attached with a header indicating the data destination on the transmission path. It is transmitted in fixed length cells of several bytes. In this case, if a ladder error or the network's processing capacity is exceeded due to too many cells, the cells will be discarded within the network and may not reach the receiving side.

In such cases, conventionally, image data is output by using the coefficient data of the corresponding block in the frame immediately before the frame containing the block corresponding to the discarded cell as the data of the discarded block. Ta.

Therefore, when cells are discarded, there is a problem in that the image deteriorates in the receiving section. Furthermore, there is a problem in that when there is a large difference in low frequency coefficients between the corresponding block in the previous frame and the block in the current frame, the deterioration in image quality is particularly significant.

The present invention compares the magnitude relationship between the corresponding low frequency coefficients of blocks in the current frame and the previous frame on the transmitting side, and when the difference is large, transmits compensation information to the receiving side, thereby eliminating the difference when cell discard occurs. The purpose is to compensate for coefficients with large values using compensation information.

[Means to solve the problem]

FIG. 1 is a block diagram of the principle configuration of an image data transmission system using the cell discard compensation image decoding method of the present invention.

In the same figure, input image data is subjected to, for example, discrete cosine transform by an orthogonal transform unit 14, and the input image data is
5 and output from the transmitter to the transmission path 16.

On the receiving section side, data input from the transmission path 16 is decoded by a decoding section 17, and subjected to, for example, inverse discrete cosine transformation by an inverse orthogonal transform section 12, and outputted as image data. In the first invention and the second invention, the content that the coefficient magnitude relation transmitting means 1 outputs to the transmission line 16 and the action of the discard compensation means 13 provided in the decoding section 17, for example, are each one part. The configurations of the block diagrams are identical except for the following points.

In Figure 1, input image data is orthogonally transformed using a plurality of pixels as a block, and the coefficients in the orthogonal transform coefficient domain are encoded and transmitted in cell units. A plurality of coefficients representing components are grouped into a plurality of groups corresponding to a plurality of same wavenumber bands.

In the first invention, the coefficient magnitude relation transmitting means 11 includes a plurality of coefficients in the group with the lowest frequency among the plurality of groups in the block currently being transmitted on the transmitting side of the image data transmission system and this block. Whether the difference between the corresponding coefficients in the corresponding block in the frame one frame before the frame is larger or smaller than the predetermined darkness value for each coefficient. is determined, and the magnitude relationship information indicating the result and the value of the coefficient in the current block for the coefficient with a large difference are used as compensation information, for example, a value that has been quantized more coarsely than for the coefficient in the block. It is transmitted to the receiving side of the system via the transmission line 16.

For example, on the receiving side of the image data transmission system, the decoding unit 17
A discard compensating means 13 provided inside detects the presence or absence of cell discard from the contents of the cell transmitted via the transmission path 16, and when a cell is discarded, transmits to the block corresponding to the discarded cell. Using the information indicating the magnitude relationship of the difference between the coefficients transmitted from the side coefficient magnitude relationship transmitting means 11,
Among the coefficients in the lowest frequency group mentioned above, for coefficients with a small difference, the corresponding coefficients in the corresponding block one frame before the frame containing that block are used, and for coefficients with a large difference, the corresponding coefficients in that block are used. The coefficients are replaced with the coefficient values transmitted from the coefficient magnitude relation transmitting means 11 and output to the inverse orthogonal transform section 12.

In the second invention, the coefficient magnitude relation transmitting means 11 on the transmitting side transmits information indicating whether the difference between the respective coefficients is larger or smaller than a predetermined darkness value, and, for coefficients with a large difference, the difference between the coefficients. Send the value of to the receiver. In response to this, the discard compensation means 13 on the receiving side adds the transmitted difference and the value of the corresponding coefficient in the corresponding block one frame before for coefficients with a large difference, and outputs the result to the inverse orthogonal transform unit 12. The same operation as in the first invention is performed except that.

[For production]

In the present invention, coefficients in the orthogonal transform coefficient domain are divided into multiple groups corresponding to multiple frequency bands, and the cell discard compensation image decoding method of the present invention is applied to the lowest frequency group among them. be done. As explained in FIG. 10, the coefficient at the top left of the coefficient area indicates the DC component, and the frequency of the AC component increases toward the bottom right. Therefore, for example, by dividing the coefficient region by a broken line extending from the upper right to the lower left, coefficients corresponding to a plurality of frequency bands are grouped.

Then, for a plurality of coefficients in the lowest frequency group among these groups, the difference is calculated between the respective corresponding coefficients in the current frame and the frame one frame before,
In the first invention, the magnitude relationship information indicating whether the difference is larger or smaller than the dark value predetermined for each coefficient, and when the difference is large, the value of the coefficient in the current block is also In the second invention, the difference value itself is sent to the receiving side, and the receiving side sends the information indicating the magnitude relationship of each coefficient sent from the sending side and the value of the coefficient of the current frame when the difference is large, or the value of the difference. When cell discard occurs, the value of the coefficient in the previous frame is compensated and output using the information shown in FIG. Further, as the coefficients of groups other than the lowest frequency group, for example, the values of the corresponding coefficients in the previous frame are output as they are.

As described above, in the present invention, even if the coefficients of a block in the current frame are lost due to cell discard, it is possible to output image data by reducing the error in low-frequency coefficients among those coefficients. .

〔Example〕

FIG. 2 is an embodiment of the method for setting frequency bands in the orthogonal transform coefficient domain according to the present invention. In the figure, the upper leftmost coefficient indicates the DC component, and all other coefficients indicate the AC component.
For the AC component, the lower right area in the diagram shows the highest frequency. Therefore, as shown in the figure, the frequency bands can be separated by broken lines (thick lines) extending from the upper right side to the lower left side of the coefficient area. Although the figure shows a coefficient area for 8×8 pixels, the number of pixels is not limited to this, and it goes without saying that the way frequency bands are divided by thick broken lines is not limited to this. It is.

FIG. 3 is a conceptual diagram of an embodiment of the first invention for coefficients of the lowest frequency group consisting of three coefficients. - In the same figure, (a) shows the coefficients in the frame one frame before the current frame, and the group with the lowest frequency includes three, Ll for the DC component, and L2 and L3 for the AC component. The figure (b) shows the coefficients in the current frame, and the lowest frequency group consists of 11 for the DC component and 12 and 13 for the AC component. Note that among the coefficients in these frames, the coefficients other than those in the lowest frequency group are not applicable to the present invention, so their values are all set to H.

By comparing the coefficients of the lowest frequency groups in FIGS. 3(a) and (c) with each other, the difference between the coefficients in the current frame and the coefficients in the previous frame is predetermined for each coefficient. It is determined whether it is larger or smaller than the darkness value. Here, the threshold values for the respective coefficients may all be the same, but different values may be used because the influence on the overall image differs depending on the frequency. For example, if the difference between the DC component 11 and Ll and the difference between 13 and L3 of the AC bone exceed the respective dark values, the coefficients 11 and 13 together with information indicating the magnitude relationship of the respective coefficients. are transmitted to the receiving side as (11) and (I13), which are results of coarser quantization than in normal coefficient transmission.

FIG. 3C shows the coefficients that are output as the current frame after normal decoding when no cell discard occurs. On the other hand, Figure (d) shows the coefficients after decoding when cell discard occurs, and among the coefficients in the lowest frequency domain, the coefficients with a small difference between the current frame and the previous frame are The value L2 of the frame is (11) and (!3) sent from the transmitter for coefficients with large differences.
is output.

FIG. 4 is a block diagram of the overall configuration of an embodiment of the image data transmission system according to the present invention. In the same figure, the transmitting side uses the output of the subtracter 22, which will be described later, for each coefficient in the lowest frequency group, as well as information indicating the magnitude relationship of the coefficients between the current frame and the previous frame. A magnitude relation and compensation information transmitter 19 transmits the value of a large current intra-frame coefficient, or in the second invention, the value of the difference between both coefficients to a discard compensator 30, which will be described later, on the receiving side, and orthogonally transforms the input image data. A DCT unit 2OSDCT unit 20 performs discrete cosine transformation (DCT), which is a type of
a subtracter 22 that takes the difference between the output ■ and the corresponding block data ■S in the previous frame stored in the frame memory 21; a quantizer 23 that quantizes the output ■S of the subtractor 22; The output of the quantizer 23 is returned to the form before quantization■
An adder 25 that adds the output of the inverse quantizer 24, S, and block data in the previous frame, which is the output of the frame memory 21, S, and an output of the adder 25, S. It consists of a leak coefficient multiplier 26 which multiplies the leak coefficient α by a leak coefficient α and outputs it to the frame memory 21.

Also, on the receiving side, the signal ■S input via the transmission path
An inverse quantizer 27 that returns the quantized input data to its original format ■R, and an addition that takes the sum of its output ■R and the block data ■R in the previous frame stored in the frame memory 28 29, a discard compensator 30 corresponding to the discard compensator 13 in FIG. Disposal compensation department 30
The frame memory 28 multiplies the output ■R by the leakage coefficient α.
It consists of a leak coefficient multiplier 32 that outputs a leak coefficient.

FIG. 4 shows an embodiment of the present invention using the interframe coding method. For example, on the transmitting side, the difference between the block data S in the frame to be currently transmitted and the block data S in the previously transmitted frame. , that is, the inter-frame difference is taken as ■S, which is quantized and, for example, Huffman encoded,
Output on the transmission path. On the receiving side, the difference ■R from the transmission path
By receiving this and adding it to the block data ■R in the previous frame, the block data ■R in the current frame is obtained. Then, on the transmitting side, the sum of the difference ■S and the block data ■S in the previous frame is added to the adder 2.
5, multiplied by the leakage coefficient α, and then stored in the frame memory 21 in order to obtain the inter-frame difference for the next frame. In the receiving section as well, the current block data (1) R is similarly multiplied by the leakage coefficient α and then stored in the frame memory 28.

For example, the leak coefficient α is used to attenuate the effect of cell discard in a short time even if cell discard occurs. For example, the difference D1 is input on the receiving side, and the frame memory 28
Block data F of the previous frame stored in
It is assumed that the sum (D++FMo) is multiplied by the leakage coefficient α, and the block data for the current frame expressed by the following equation becomes the content FM1 of the frame memory 28.

FM+ = (D+ +FMo) α= DI α+F
Moα Then, when the difference D2, which is block data between the next frames, is input through the transmission line, the contents of the frame memory 28 become as follows.

F Mz = (Dz + F M+ )α=Dzα+
D+α+FMo α2 Furthermore, when block data D3 between the next frames is input, the contents of the frame memory 28 are given by the following equation.

FM27- (02+FM2)α=D3α+D2
α2+D, α”0FM6a3 For example, the prediction error DI when cell discard occurs
The influence on the entire image attenuates to α7 after inputting n frames. Therefore, by setting α smaller than 1, the influence can be reduced to almost zero in a short time.

Note that the discrete cosine transformation (DC
T) performs orthogonal transformation of image data using a transformation matrix made from a cosine function, and it required a multiplier for calculation, making it difficult to perform high-speed calculations, but with the development of high-speed signal processing processors It has become popular for semi-video transmission such as still image transmission and video conference signals.

FIG. 5 is an example of coefficient data of the lowest frequency group in the image data transmission system of FIG. 4. This figure shows an example of data in the first invention, in which when the difference between the coefficient values is large, the coefficient values themselves in the current frame are roughly quantized and transmitted from the transmitting side. In the figure, only the coefficients in the lowest frequency group are shown, and all coefficients indicating other high frequency components are omitted.

Among the data on the transmitting side in FIG. 5(a), ■S indicates the coefficients of the lowest frequency group for six consecutive blocks in the current frame. ■S indicates the coefficient of the lowest frequency group in the previous frame, and the difference between ■S and ■S is quantized by the subtracter 22 in FIG. 4, which indicates the interframe signal. The value is encoded and
Output on the transmission path.

Here, the threshold value for determining the size between the data ■S of the current frame and the data ■S in the previous frame is set to the same 8 for each coefficient, and when the difference is larger than the dark value, it is marked with the symbol l, and it is If the case is represented by 0, the magnitude relationship in FIG. 4 and the compensation information outputted from the compensation information transmitter 19 onto the transmission path are as follows.

ooo ooo ooo.

1 (compensation information for 30) 01 (compensation information for 1) ooo
oo.

The vertical bars of this auxiliary information indicate the separation of six consecutive blocks (1st to 3rd blocks from the left in the upper row, 4th to 6th blocks from the left in the lower row), and the first to third blocks, For the fifth and sixth blocks, 0 is output for all coefficients, indicating that the difference between the corresponding coefficients is smaller than the dark value.

On the other hand, for the fourth block, auxiliary information (for example, 3
2) is shown in parentheses, then the code 0 indicates that the difference between the coefficients on the right side of the DC component is smaller than the dark value, and then the difference between the coefficients on the lower side of the DC component -9 is less than the dark value. Following the code 1 indicating that it is large, auxiliary information (
For example, O) is shown in parentheses.

Figures 5 and 5) are examples of coefficient data on the receiving side. In the figure, the coefficient ■R for the previous frame is equal to the data ■S for the previous frame on the transmitting side. Then, as shown in R, if it is assumed that the cells for the fourth and fifth blocks among the cells input from the transmission path are discarded, then the above-mentioned size relationship and the compensation information transmitter 1
Using the auxiliary information from 9, the contents of ■R to be output to the inverse discrete cosine transformer 31 in FIG. 4 are created.

That is, for the fifth block of (1)R, all coefficient values of (2)R are used as they are. On the other hand, the fourth
For the block, the value 32 of the auxiliary information sent from the transmitting side as the DC component, and the value 0 also sent from the transmitting side as the coefficient below it, and for the coefficient on the right side of the DC component, The value 13 in the data ■R in the previous frame is used as the data for the current frame.

FIG. 6 shows the second invention, that is, when the difference between the coefficients of the current frame and the previous frame is large, the difference is coarsely quantized and sent to the receiving side. This is an example. In the figure, the size relationship on the transmitting side and the auxiliary information sent from the auxiliary information transmitter 19 to the receiving side are as follows.

+00010001000 +1 (22 compensation information) 01 (-9 compensation information) 1oo
ooo.

That is, as auxiliary information for the difference 22 to the DC component of the fourth block, a value obtained by quantizing 22, for example 24, is sent to the receiving side, and as auxiliary information for the difference -9 to the AC component below the DC component, for example -8 is sent to the receiving side. The DC component of the fourth block of R that is sent and input to the inverse discrete cosine transformer 31 on the receiving side is 32, the AC component to the right thereof is 13, and the AC component below the DC component is 2. The block data other than the ``R'' group on the receiving side, ``R1'' and the block data on the transmitting side are all the same as in FIG.

FIG. 7 is an embodiment of the magnitude relationship and the processing flowchart of the compensation information transmitter 19 in the first invention. In the figure, when the process starts, it is first determined in step 33 whether or not the processing for the lowest frequency group has been completed for all coefficients. If not, in step 34, the coefficient value of the current frame and the coefficient value of the previous frame are determined. The difference with the coefficient value is calculated. Then, in step 35, it is determined whether the difference is larger than the dark value, and if the difference is small, in step 36, the fact that the difference is small is transmitted to the receiving side, and the process returns to step 33.

If the difference is greater than the dark value in step 35, the fact that the difference is greater is transmitted to the receiver in step 37, and at the same time,
In step 38, the coefficient values in the current frame are coarsely quantized and transmitted, and the process returns to step 33. The process from step 33 is repeated for all the coefficients of the lowest frequency group, and the process ends when all the coefficients have been processed.

In the second invention, when the difference is larger than the dark value in step 35, the fact that the difference is large is transmitted in step 37, and at the same time, in step 38, the difference is coarsely quantized, rather than the coefficient value itself of the current frame. The process is exactly the same, except that it is sent as

FIG. 8 is a flowchart of a processing example of the discard compensation unit 30. In the figure, when the process starts, first in step 39 it is determined whether or not cells are to be discarded. If there is no discard, the discard compensation section 30 outputs the output from the adder 29 as it is to the inverse DCT section 31 without performing any processing.

If a cell is discarded, the coefficients in the previous frame are replaced for low frequencies for which there is a large difference between the coefficients of the current frame and the previous frame, and the process ends. Here, in the first invention, the coefficient replacement is a value obtained by quantizing the coefficient value of the current frame □ sent from the receiving □ side,
Further, in the second invention, a value obtained by quantizing the difference between coefficients is used.

In the above explanation, it is assumed that coefficients with large differences or quantization of the differences are performed more coarsely than normal quantization by the quantizer 23 in FIG. 4, but this is only used when auxiliary information is discarded as a cell. Therefore, the quantization is performed by the quantizer 23
Of course, it is okay to do it at the same level as. Furthermore, although the present invention has been described in detail using an interframe encoding method as an example, the present invention is not limited to this encoding method, and the present invention can be applied even to intraframe encoding.

〔Effect of the invention〕

As described in detail above, according to the present invention, even if cells are discarded, it is possible to reduce the error in the low frequency coefficient of the current frame, and the effect on image quality when cells are discarded can be reduced. .

[Brief explanation of drawings]

Figure 1 is a block diagram showing the principle configuration of the image data transmission system according to the present invention, Figure 2 is a diagram showing an example of a method for setting frequency bands in the coefficient domain, and Figure 3 is the lowest frequency consisting of three coefficients. Figure 4 is a block diagram showing the overall configuration of the embodiment of the image data transmission system; Figure 5 is the coefficient of the lowest frequency group in the first invention. FIG. 6 is a diagram showing an example of coefficient data compensated on the receiving side in the second invention; FIG. 7 is a diagram showing the magnitude relationship and processing of the compensation information transmitting unit in the first invention FIG. 8 is a diagram showing a flowchart of the processing example of the discard compensation section. FIG. 9 is a block diagram showing the overall configuration of an image data transmission system using orthonormal transformation. FIG. FIG. 3 is a diagram showing a transform coefficient area after orthogonal transform in an image. 12... Inverse orthogonal transformation section, 14... Orthogonal transformation section, 15... Encoding section, 16... Transmission section, 17... Decoding section, 19... Size relationship and compensation information transmission Part, 20...
Discrete cosine transform (DCT) unit, 21.28... Frame memory, 30... Discard compensation unit, 31... Inverse DCT unit.

Claims (1)

[Claims] 1) An orthogonal transform unit (14) orthogonally transforms the input image data using a plurality of pixels as a block, encodes the coefficients in the orthogonal transform coefficient region, transmits the code in cell units, and inverts the input image data to the receiving side. In an image data transmission system having an inverse orthogonal transform unit (12) that performs orthogonal transform, a plurality of coefficients representing frequency components from direct current to high frequencies in the orthogonal transform coefficient region are grouped into a plurality of groups corresponding to a plurality of frequency bands. and transmits to the transmitting side of the image data transmission system, a plurality of coefficients in the group with the lowest frequency among the plurality of groups in the block currently being transmitted, and a frame temporally one frame before the frame including the block. information indicating whether the difference between each of the coefficients in the block corresponding to the block is large or small, and the current state of the coefficient with the large difference. A coefficient magnitude relation transmitting means (11) for transmitting the value of the coefficient in the block to the receiving side of the first half image data transmission system, and a coefficient size relation transmitting means (11) for detecting the presence or absence of cell discard to the first half image data transmission system receiving side, and detecting whether or not cells are discarded. If there is, the coefficient magnitude relation transmitting means (11) for the block corresponding to the discarded cell.
Based on the information indicating the magnitude relationship of the difference between the coefficients transmitted from The inverse orthogonal transform unit (11) replaces the corresponding coefficient data, or for coefficients with a large difference, the data in which the corresponding coefficient in the block is replaced with the coefficient value transmitted from the coefficient magnitude relationship transmitting means (11).
2) A cell discard compensation image decoding system characterized by comprising a discard compensation means (13) for outputting. 2) Inverse orthogonal processing in which the input image data is orthogonally transformed using a plurality of pixels as a block by the orthogonal transform unit (14), the coefficients in the orthogonal transform coefficient region are encoded and transmitted in units of cells, and the inverse orthogonal transform is performed on the receiving side. In an image data transmission system having a transform unit (12), a plurality of coefficients representing frequency components from direct current to high frequencies in the orthogonal transform coefficient region are grouped into a plurality of groups corresponding to a plurality of frequency bands, and the image data transmission The transmission side of the system is provided with a plurality of coefficients in the group with the lowest frequency among the plurality of groups in the block currently being transmitted, and a plurality of coefficients corresponding to the block in a frame temporally one frame before the frame including the block. information indicating whether the difference between the respective coefficients is large or small between the plurality of coefficients respectively corresponding to the plurality of coefficients in the block, and the difference value of the coefficient with respect to the coefficient with the large difference. Coefficient magnitude relationship transmitting means (11) for transmitting to the receiving side of the image data transmission system
and, on the receiving side of the image data transmission system, detecting the presence or absence of discarded cells, and when there is discarded cell, transmitting means (11) for transmitting the coefficient magnitude relation to the block corresponding to the discarded cell.
Based on the information indicating the magnitude relationship of the difference between the coefficients transmitted from Using the corresponding coefficients, and for coefficients with a large difference, the coefficient difference value transmitted from the coefficient magnitude relationship transmission means (11) and the block corresponding to the block one frame before the frame containing the block. A cell discard compensation image decoding system, comprising a discard compensating means (13) for outputting coefficients in the lowest frequency group to an inverse orthogonal transform unit (12) using corresponding coefficients in the lowest frequency group.