JPH043684A - Variable rate moving image encoder - Google Patents

Abstract

PURPOSE:To suppress the deterioration of picture quality to a low level by evaluating each block in a predictive error signal region or a discrete cosine transformation coefficient region, attaching priority on each block unit according to a certain rule, and selecting a variable speed channel corresponding to the priority. CONSTITUTION:A priority decision means 13 which attaches the priority on each block according to a certain rule by using the transformation coefficients in which a predictive error signal value or a predictive error signal in each block is transformed at a discrete cosine circuit 11, and selects the variable speed channels 14, 15 corresponding to attached priority is provided. Therefore, only a unique block (on which low priority is attached) not being prominent visually even when it is deleted is targeted to be deleted when a network is set in a congested state, and the unique block (on which high priority is attached) to generate remarkable deterioration visually when it is deleted is not deleted since it is transmitted on a variable speed high priority channel 14. In such a way, it is possible to suppress the deterioration of the picture quality to a low level even when the deletion of a packet due to the congestion of the network occurs.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a variable rate moving image encoding device used in video telephones, video conference systems, and the like.

BACKGROUND OF THE INVENTION FIG. 2 shows a schematic configuration of an example of a conventional variable rate video encoding device. In FIG. 2, moving image signal 2
0 is manually input in frame units. On the other hand, the predicted signal of the current frame is obtained by subjecting the previous frame reproduced signal stored in the frame memory 28 to motion compensation or the like. The obtained prediction signal and the moving image signal 20 are subtracted to obtain a prediction error signal, which is then processed for each block of size N×N.

The N×N prediction error signal is subjected to discrete cosine transform in the discrete cosine transform circuit 21 and then input to the quantization circuit 22 . The quantization circuit 22 quantizes the discrete cosine transformed coefficients, and the quantized representative value is input to the hierarchization circuit 23 . The hierarchization circuit 23 converts the N×N quantized representative value into a portion of the prediction error signal that includes low frequency components (hereinafter referred to as MSP).
It is abbreviated as ) and the part containing high frequency components of the prediction error signal (
Hereinafter, it will be abbreviated as LSP. ), MSP is transmitted on a variable rate high priority channel 24 and LSP is transmitted on a variable rate low priority channel 25. The variable speed high priority channel 24 and the variable speed low priority channel 25 are both variable speed channels that can support any transmission speed, and are realized by wideband I5DN, which is currently being researched as a next generation communication network.

In broadband 15DN, multiple variable speed channels with different priorities will be prepared, and when the communication network (hereinafter referred to as the network) becomes overcrowded, control will be implemented such as discarding packets from channels with lower priority. It is. Therefore, the variable speed high priority channel 24 is a channel with high priority, i.e., packets are not discarded even when the network is congested, and the variable speed low priority channel 25 is a channel where packets are discarded when the network is congested. This is a channel that may be used.

Local decoding is performed using only the MSP, and decoding on the receiving side is performed by adding the MSP to the value of the previous frame and then adding the decoded signal of the LSP.

Therefore, when the network is busy, LSPs are probabilistically discarded, but since the LSPs themselves have little influence on image quality and do not spread over time, deterioration in image quality can be kept to a minimum.

Problems to be Solved by the Invention Generally, when a network is busy, there is a high possibility that a plurality of consecutive packets will be discarded. However, in the conventional example, without considering the characteristics of the significant block (a block in which at least one quantization representative value that is not O out of N×N quantization representative values exists), all significant blocks are Since the MSP and LSP are separated and transmitted according to a certain rule, there is a possibility that LSPs of all significant blocks may be discarded. For example, normally, many significant blocks tend to occur in a dynamic region, but in the conventional example, it is possible that LSPs of all significant blocks generated in a dynamic region are discarded.

The present invention solves the above problem, and instead of transmitting significant blocks separately, a priority is assigned to each significant block according to a certain rule, and the transmission is performed on a block-by-block basis according to the assigned priority. An object of the present invention is to provide a variable rate video encoding device that selects a channel according to the timing.

Means for Solving the Problems In order to achieve the above object, the present invention evaluates each block in the prediction error signal domain or the discrete cosine transform coefficient domain, and assigns priority to each block according to a certain rule. The variable speed channel is selected according to the priority level.

Effects of the present invention With the above configuration, when the network is congested, only significant flocs that are not visually noticeable even if they are discarded (significant blocks given a low priority) are subject to discarding. Significant blocks that cause significant visual deterioration (significant blocks given high priority)
are not discarded for transmission on variable rate high priority channels. Therefore, even if packets are discarded due to network congestion, image quality deterioration can be kept to a minimum.

EXAMPLES Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a variable rate video encoding device according to the present invention. In FIG. 1, 10 is a moving image signal, 11 is a discrete cosine transform circuit that performs discrete cosine transform on the interframe prediction error signal in units of N×N blocks, and 12 is a coefficient transformed by the discrete cosine transform circuit 11. This is a quantization circuit that quantizes. Reference numeral 13 denotes a priority determination circuit that determines the priority of each block using the prediction error signal value within the block or using the transform coefficient obtained by performing discrete cosine transform on the prediction error signal. 14 and 15
Both are variable speed channels that can support arbitrary transmission speeds, 14 is a variable speed high priority channel that has a high priority, that is, packets are not discarded even when the network is busy, and 15 is a variable speed high priority channel that can support any transmission speed. It is a variable rate low priority channel where packets may be discarded at times.

16 and 17 are an inverse quantization circuit and an inverse discrete cosine transform circuit for local decoding. 18 is a frame memory for storing reproduced images, and 19 is a prediction coefficient multiplier.

Next, the operation of the above embodiment will be explained. In FIG. 1, a moving image signal 10 is input frame by frame. On the other hand, the predicted signal of the current frame is obtained by applying motion compensation etc. to the previous frame reproduced signal stored in the frame memory 18.
It is obtained by multiplying the prediction coefficient a (O × α≦1) by the prediction coefficient multiplier 19. The obtained prediction signal and the moving image signal 10 are subtracted to become a prediction error signal, which is hereinafter N×
Processing is performed for each block of size N.

The N×N prediction error signal is subjected to discrete cosine transform in the discrete cosine transform circuit 11 and then input to the quantization circuit 12 . The quantization circuit 12 quantizes the coefficients subjected to the discrete cosine transform, and the quantization representative value is sent to the priority determination circuit 1.
3 and is also input to an inverse quantization circuit 16 for local decoding. The coefficients dequantized by the dequantization circuit 16 are subjected to inverse discrete cosine transform in the inverse discrete cosine transform circuit 17, and then added to the prediction signal and stored in the frame memory 18.

The priority determination circuit 13 evaluates the prediction error signal on a block-by-block basis and determines the priority of the block, and blocks given high priority are assigned to the variable speed high priority channel 14.
In addition, blocks given lower priority are transmitted on the variable rate low priority channel 15, respectively.

Next, a method for determining block priorities will be described using several examples.

The size of a block serving as a processing unit is N×N, and the input video signal in a certain block is x (m, n), the predicted signal is x (m, n), and the prediction error signal is d (m, n). However, m=o, 1.2. , N-1 n"Or 1.2.-, N-1, and cl(m, n)=x(m, n)-x(m, n).

[Method 1] Method using visual discrimination threshold First, from the following equation, Δ
. Find v (, compare the obtained Δ to νC with the threshold Δtk value determined in advance from the visual discrimination threshold, etc., and if Δav6 ≧ Δtl, select the high priority block Δave
If <Δth, the block is considered a low priority block.

Alternatively, let η be the number of Δ(m, n) that satisfies Δ(m, n)≧Δth in the block, compare this with 9 preset thresholds, and find that η≧ ηt1. l, then high priority block ηt
h, it is considered a low priority block.

[Method 2] Method using mean square error First, find MSE from the following formula, N×N m=U n=[J Compare the found MSE with a preset threshold MSEth, and if MSE≧MSEth, give high priority. degree block MSE<
If it is MSEth, it is considered a low priority block.

In this way, according to the embodiment, when the network is busy, only packets made up of low-priority significant blocks are discarded, and packets made up of high-priority significant blocks are not discarded. Furthermore, when a packet consisting of low-priority significant blocks is discarded, d(m) is added to the image data during transmission and reception.

n), but this effect becomes d(m,n)xα1 after k frames. That is, the prediction coefficient a is a parameter that determines the recovery time after packet discard. Therefore, by appropriately selecting the significant block priority determination method and the prediction coefficient a, it is possible to make the image quality deterioration not visually noticeable even when packets are discarded.

In addition, in the above embodiment, two methods using prediction error signal values were shown as methods for determining the priority of significant blocks.
The present invention is not limited to these two methods.

As described in detail, the present invention evaluates each block in a prediction error signal domain or a discrete cosine transform coefficient domain, assigns a priority to each block according to a certain rule, and performs a process according to the assigned priority. Since variable rate channel selection is performed, when the network is congested, only packets consisting of low-priority significant blocks are discarded, and packets consisting of high-priority significant blocks are discarded. Therefore, even if packets are discarded, image quality deterioration can be kept to a minimum. In addition, even if a packet consisting of low-priority significant blocks is discarded, by appropriately selecting the method for determining the priority of significant blocks and the prediction coefficient α, we can prevent the deterioration of image quality from being visually noticeable. can do.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of a variable rate video encoding device according to the present invention, and FIG. 2 is a block diagram showing an example of a conventional variable rate video encoding device. DESCRIPTION OF SYMBOLS 10... Video signal, 11... Discrete cosine transform circuit, 12... Quantization circuit, 13... Priority determination circuit, 14... Variable speed high priority channel, 15... Variable Speed low priority channel, 16... Inverse quantization circuit, 1
7... Inverse discrete cosine transform circuit, 18... Frame memory, 19... Prediction coefficient multiplier.

Claims (1)

[Claims] In interframe predictive coding of video signals in which a prediction error signal is divided into blocks of N×N size and processed, prediction error signal values in each block or transform coefficients obtained by discrete cosine transformation of the prediction error signal are used. A variable rate video encoding device comprising priority determining means for assigning a priority to each block according to a certain rule and selecting a variable rate channel according to the assigned priority.