JPH04120888A - Moving image encoder - Google Patents

Abstract

PURPOSE:To hardly receive degradation against cell abandonment by directly encoding a composite signal and further allocating the information of one block within one cell. CONSTITUTION:This device is composed of a motion compensator 1, significant block decider 2, discrete cosine transformer 3, quantizer 4, information limiter 5, inverse quantizer 6, inverse discrete cosine transformer 7 and frame memory 8. The composite signal is used for an input picture signal to be directly encoded, and the information limiter 5 limits the information amount of a quantized block element so as to allocate the information of one block within one cell. Thus, degradation is hardly received against cell abandonment.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a video encoding device suitable for an asynchronous transfer mode (hereinafter referred to as ATM), and in particular a video encoding device that is less susceptible to deterioration due to cell discard. Regarding.

[Prior Art] FIGS. 2 and 3 show a conventional high-efficiency video encoding device.

In FIG. 4, the composite signal is transferred to a Y/C separator 10.
1 and separated C1 and C2 signals (color difference signals)
is subsampled, and the Y signal (luminance signal), C1 signal, and C2 signal are each encoded by encoding devices 102, 103, and 104.
, and quantization information is output from this. Each of the encoding devices 102.1, 3.104 is a high-efficiency encoding device for maintaining uniform image quality, and has a block configuration as shown in FIG.

In the encoding apparatus shown in FIG. 3, the difference between the block of the previous frame processed by the motion compensator 1.5 and the block of the current frame is calculated, and the significant block discriminator 106 determines the significant block for the difference block. If the block is determined to be a significant block, a discrete cosine transformer 107 applies a discrete cosine transform (hereinafter referred to as DCT), a quantizer 108 performs quantization, and then transmission is performed. Furthermore, the quantized information that has been quantized at the same time is inversely quantized by an inverse quantizer 109, and inversely discrete cosine transform (hereinafter referred to as IDCT) is applied by an inverse discrete cosine transformer 110, and based on the difference signal, put it back,
It is stored in a frame memory (hereinafter referred to as FM) 111. In addition,
The block referred to here is a block defined for movement measurement.
A dimensional small block is an input image signal divided into several blocks. For example, 8×8
Alternatively, it is composed of 16×16 pixels.

By the way, the above-mentioned quantized information is sent to a high-speed ATM transmission line, but the ATM system incorporates the advantages of the packet communication system, dividing the information to be transmitted into small packets (called cells), and each divided Transfer cells by fitting them into empty time slots. A header indicating the destination etc. is attached to the beginning of each cell, and within the ATM network, the content of the header is referred to and processing is performed on a cell-by-cell basis.

In the packet communication system, there is no concept of discard, whereas in the ATM system, when the total number of cells exceeds the processing capacity of the network due to network congestion, a phenomenon called cell discard occurs. When cell discard occurs, transmission quality deteriorates, such as part of information being lost. Therefore, especially in the case of image communication, deterioration of image quality occurs, such as loss of part of the image.

[Problem to be solved by the invention] When a conventional encoding device that performs Y/C separation and quantization before transmission is used in an ATM network, there is a problem that it becomes susceptible to deterioration due to cell discard. There is.

For example, if many Y signals are included in the discarded cells, a large amount of information will be lost and the discarded Y signals will be lost.
When the C1 signal and 02 signal corresponding to the signal are transmitted, there is a problem in that there is a high possibility that these pieces of information will be meaningless due to the absence of the Y signal.

This will be specifically explained using FIG. 4. An ATM cell consists of a header and an information section called a payload. The CCI TT Ordinance has issued a draft that states that it would be better to be able to completely combine signals within one cell.
- Boat fishing is not designed to be completed within one cell.

For this reason, normally the block information BLKI as shown in the figure
, BLK2, . . . do not fit within each cell 1, cell 2, but are embedded across cells 1, 2, and 3.

For this reason, for example, if cell 1 containing part of BLKI is discarded after transmission, even if cell 2 is valid, the remaining BLKI data in it is incomplete and useless data is sent. It turns out to be a problem.

An object of the present invention is to directly encode a composite signal, in order to eliminate the above-mentioned problem of being susceptible to deterioration due to cell discard in an ATM network.
Furthermore, by allocating one block of information in one cell, the present invention aims to provide an image encoding device that is less susceptible to deterioration due to cell discard in an ATM network.

[Means for Solving the Problems] This invention generates a predicted value using an input image signal to be encoded, subtracts the generated predicted value from the input image signal to obtain a prediction error, and calculates a prediction error from this prediction error. Determine a significant block, perform orthogonal transformation on the significant block, scan the transformed block in a predetermined direction, discard block elements that are below a threshold, and quantize block elements that exceed the threshold to create cells. It is applied to an image encoding device that transmits units to an ATM network.

In such an image encoding device, the input image signal is directly encoded using a composite signal, and the amount of information of the quantized block element is limited, so that 1
An information limiter is provided to allocate one block of information within a cell.

[Effect] When the input image signal is directly encoded using a composite signal, the ATM network processes the integrated information without Y/C separation, so even if cells are discarded, the ATM network When the C1 and C2 signals corresponding to the discarded Y signal are transmitted, as in the case of a separate coding method in which Y/C is separated and processed, these information is lost due to the absence of the Y signal. There is no such thing as becoming meaningful.

In addition, if an information limiter is provided to limit the amount of information of a quantized block element and one block of information is allocated within one cell, one block of information is completed with the information within one cell. Even if a certain cell is discarded, the block information in the discarded cell remains within the discarded cell and does not affect other valid cells that are not discarded, eliminating unnecessary transmission.

[Example code] Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows an embodiment of a moving picture encoding device according to the present invention. In the moving picture encoding device in an ATM network, a composite signal is used as an input image signal, and one block of information is stored in one cell. In order to allocate the block elements, a means is provided to limit the number of block elements to be quantized.

The motion compensator 1 generates motion vector information from an input screen block consisting of a composite signal and a previous screen block from the frame memory 8, and a block obtained by moving the previous screen block according to this motion vector information is defined as a predicted screen block, The prediction error between this and the input screen block is taken. This prediction error is added to the significant block determiner 2,
When the significant block is determined to be a significant block by the significant block determiner 2, the significant block information is subjected to a discrete cosine transform by the discrete cosine transformer 3, and then added to the quantizer 4. In the quantizer 4, block elements are searched in the zigzag scan direction shown in FIG. 6, block elements below the threshold are discarded, run-length codes are used for the discarded ones, and block elements exceeding the threshold are searched for. Quantization is applied only to

This quantized information is not transmitted as is, but is first input to the information limiter 5. This information limiter 5 has a function of limiting the amount of information per block to within one cell.

Then, after the amount of information per block is limited, it is sent out to a high-speed transmission path.

FIG. 7 shows a processing flow explaining the function of the information limiter 50 described above, the contents of which will be explained below. Here, the cell length sent out to the transmission path is cl [bytes], the number of quantization bits is q [bits], the number of bits used for run length is RL [bits], and the zigzag scan up to the present time (see Figure 6) is Information byte b [b
ytesco.

The information limiter 5 first recognizes the beginning of the block element, and then clears the information byte b (step 301).
, 302). Next, it is determined whether there is an output from the quantizer 4 (step 303), and when there is an output from the quantizer 4, the number of quantized bits q is counted, and when there is no output from the quantizer 4, the run length is counted. The number of bits RL used for the code is counted and added to each information byte b (steps 304 and 307). Then, it is determined whether or not the information b) exceeds the cell length C1 (step 305), and when the information b) exceeds the cell length cl, one block of information is completed (step 306) and one cell is The task is to correlate that information.

FIG. 5 shows a cell configuration in which decoding is completed within one cell according to the present embodiment described above. Cell 1° Cell 2. Each block information BLKI is in cell 3.

BLK2. BLK3... are stored in each cell, so even if cell 1 is discarded after transmission, the data of cell 2 following cell 1 is not affected, except that interpolation processing is required due to cell discard. This is effective for decoding BLK2 information within cell 2 without being influenced by the above, and unnecessary transmission is eliminated.

Here, this embodiment deals with moving images rather than still images. Therefore, the amount of information per block (before the limit) is 1
It is considered unlikely that the number of cells will be significantly less than the number of cells (usually 48 bytes), and in this sense, the present invention is effective.

As described above, according to this embodiment, the video encoding device in the ATM network directly encodes the input image signal using the composite signal, and in the ATM network, Y/
Since processing is performed using integrated information that is not separated by C, even if cells are discarded, it will not be possible to discard the cells, as in the case of separate coding methods, which are processed with Y/C separated in the ATM network. , c1 signal for discarded Y signal, c2
When the signals are transmitted, the information will no longer be meaningless due to the absence of the Y signal.

Additionally, since a means for limiting the number of block elements to be quantized is provided to allocate one block of information within one cell, an excellent effect of being less susceptible to deterioration due to cell discard can be expected.

Through simulations, it has been confirmed that this effect has little deterioration in image quality if the compression rate is moderate (10 Mbps or more) and the video image is suitable.

Furthermore, the present invention can also be applied to a layered encoding device by performing layering using the size of the limit amount of the information limiter. Even when a Y/C signal is used as an image signal, the present invention is applicable as long as one cell can be associated with one block using an appropriate method.

In the above embodiment, a video encoding device has been described, but this is because, as mentioned above, it is especially effective for video images.On the other hand, when applied to still images, its effectiveness is demonstrated. This is because if one block of information is allocated to one cell based on the amount of information, an empty space will be created within one cell, making it impossible to use the transmission path effectively. In addition, in the case of a still image, the present invention may be applied by allocating two or more pieces of block information within one cell.

Furthermore, in the above embodiments, the case where a discrete cosine transform, which is one of the orthogonal transforms, is used as the transform method is explained, and the effectiveness of this transform is actually confirmed by simulation. However, in principle, other methods can be applied as long as they are orthogonal transforms. For example LOT
(Lapped-Orthogonal-Trans
form) can also be used.

Note that although the present invention is limited to the ATM system, this system is not very effective in the case of a normal packet communication system because cells do not necessarily have a fixed length. However,
Even in the case of a normal packet communication system, as long as relatively short fixed length packets are used, the present invention can be applied in cases such as when an error occurs on a transmission path.

In general, the above embodiment employs a zigzag scan method to search for block elements, which is the most suitable for run-length encoding related to compression among the currently known scan methods. The direction of the scanning method is from the top left of the block containing more important information,
This is because it is directed toward the bottom right of the block containing less important information, so it is most appropriate when truncating information with an information limiter. However, it is also possible to adopt a scanning method other than the zigzag scanning method, as long as scanning with the same purpose is possible.

Further, regarding cell discard, it is suggested that the cell is discarded by the system, but in ATM, there is also discard due to delay in addition to this, and this is discarded by the user, but the present invention It can also be applied in the case of disposal.

[Effects of the Invention] According to the present invention, in a moving picture encoding device, a composite signal is used as an input image signal, and a means for limiting the number of block elements to be quantized is provided, so that one block of information is stored in one cell. Since the cell is allocated, it exhibits an excellent effect of being less susceptible to deterioration due to cell discard.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a video encoding device according to an embodiment of the present invention, FIG. 2 is a block diagram of a conventional encoding device, and FIG. 3 is a block diagram of a conventional encoding device.
4 is an explanatory diagram of conventional cell discard, FIG. 5 is an explanatory diagram of cell discard according to this embodiment, FIG. 6 is an explanatory diagram of zigzag scan, and FIG. The figure is a processing flow diagram illustrating the function of the information limiter according to this embodiment. 1... Motion compensator, 2... Significant block determiner, 3
...Discrete cosine transformer, 4...Quantizer, 5...
- Information limiter, 6... Inverse quantizer, 7... Inverse discrete cosine transformer, 8... Frame memory.

Claims (1)

[Claims] Generate a predicted value using an input image signal to be encoded, subtract the generated predicted value from the input image signal to obtain a prediction error, determine a significant block from this prediction error, In an image encoding device that orthogonally transforms the significant block and quantizes the transformed block, the input image signal is directly encoded into a composite signal, and the quantized block elements are By limiting the amount of information, 1
A video encoding device characterized in that one block of information is allocated within a cell, and this one block of information is transmitted on a cell-by-cell basis.