JPH05122685A - Encoding device between dynamic picture frames - Google Patents

Abstract

PURPOSE:To simplify the circuit constitution by sharing a motion compensation circuit for discriminating priority and non priority blocks in the encoding device between dynamic picture frames using a motion compensation forecasting. CONSTITUTION:A motion compensation circuit 5 outputs vector information detected for each block with the prescribed size for an input picture signal and a motion compensation forecasting signal and compares the motion vector with the prescribed vector threshold value to discriminate the block as a priority block. Under the threshold value the block is judged to be a non-priority block ad a priority/n-priority block control signal representing the judgement result is outputted. A packetizer 9 allocates a priority block signal to the priority cell according to the priority/non-priority block control signal, and allocates a non-priority block signal to the non-priority block. A frame memory 16 stores a local decoding signal decoding the only estimated error signal of the priority block.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an interframe coding device, and more particularly to an asynchronous transfer mode.
r Mode, hereinafter referred to as ATM. The present invention relates to a moving picture interframe coding apparatus suitable for use in a comprehensive service digital network (hereinafter referred to as ATM network).

[0002]

2. Description of the Related Art Generally, an inter-frame coding method for image signals takes a difference between an input image signal and an image signal one frame before on a transmitting side and transmits the difference signal to a receiving side.
On the receiving side, the image is reproduced by adding the difference signal sent and the signal of one frame before stored in the frame memory.

Here, since there is a still area where the inter-frame difference is zero, it is not necessary to transmit the signal to the receiving side again, and by repeatedly using the signal of the preceding frame stored in the receiving side, the image is reproduced. Can be played.

Therefore, information is generated only in a moving area, and this method of information generation is bursty.

On the other hand, for example, the ATM network is a system in which packets are assembled and transmitted each time a request for information transmission from a terminal is transmitted, and it has good compatibility with the interframe coding system.

However, even in this ATM network, there is an upper limit to the transmission speed, and if the information amount exceeding the transmission capacity is sent from each terminal all at once, the transmission cannot be completed and there is a risk that the packet will be discarded in cell units. is there. Therefore, a measure is taken to give priority to information, assign important information to priority cells, and assign other information to non-priority cells. From such a viewpoint, an interframe coding method for an ATM network that considers cell discard has been proposed.

[0007]

A conventional technique of this type is disclosed in, for example, Japanese Patent Application No. 3-53880 (Japanese Patent Laid-Open No. Hei.
There is a "moving image interframe coding device".

This method utilizes the fact that a signal having a large motion is more important in terms of image quality than a signal having a small motion among the moving image signals, and the inter-frame difference signal has a predetermined threshold. A block having a larger value is assigned to a priority cell as a priority block, and a block having a smaller value is assigned to a non-priority cell as a non-priority block.

In the above prior art, a priority / non-priority block determination circuit for determining whether the block being encoded is a priority block or a non-priority block is specially provided.

Therefore, there is a problem that the circuit configuration becomes complicated.

According to the present invention, in a moving picture interframe coding apparatus using motion-compensated prediction, a motion compensation circuit conventionally installed for motion compensation is also used for priority / non-priority block determination, so that the circuit It is an object of the present invention to provide a moving picture interframe coding apparatus having a simple configuration.

[0012]

In order to achieve the above object, the present invention has a frame memory for storing past images, and uses an output of the frame memory to input an image to be encoded. Detect the motion vector of the signal,
The detected motion is compensated to form a prediction signal, a prediction error signal formed as a difference between the input image signal and the prediction signal is encoded by a predetermined method, and the encoded data having a high degree of importance is encoded. In the moving picture interframe coding apparatus, which divides into a component and a low component, assigns a high importance component to a priority cell and assigns a low importance component to a non-priority cell, and transmits the predetermined input image signal. A motion vector is detected for each block having a size of, and the detected motion vector information and a motion compensation prediction signal are output, and the motion vector is compared with a predetermined vector threshold value, and if the threshold value is exceeded, the block Is determined as the priority block,
If it is less than the threshold value, the block is determined to be a non-priority block, and a motion compensation circuit that outputs a priority / non-priority block control signal indicating the determination result, and a priority block according to the priority / non-priority block control signal A packetizer for allocating the signal of 1 to a priority cell and a signal of a non-priority block to a non-priority block, and a first frame memory for storing a locally decoded signal obtained by decoding prediction error signals of all blocks,
A second frame memory for storing a locally decoded signal obtained by decoding only the prediction error signal of the priority block is provided.

[0013]

According to the present invention, the motion vector detected by the motion compensation circuit is used to determine whether the relevant block being encoded is a priority block or a non-priority block. Therefore, it is possible to realize with a simple circuit configuration without causing a problem that the circuit configuration becomes complicated because it is not necessary to install a special circuit.

Embodiments will be described in detail below with reference to the drawings.

[0015]

FIG. 1 shows an embodiment of the present invention. 1 is an image input terminal, 2 is a difference circuit, 3 is a valid / invalid block determination circuit, 4 is an encoding / quantization circuit, 5 is a motion compensation circuit, 6 is a multiplexing circuit, 7 is an entropy coding circuit, 8 is Buffer memory, 9 is a packetizer, 10 is a non-priority packet output terminal, 11 is a priority packet output terminal, 12 is a refresh request signal input terminal, 13 is an inverse quantization / decoding circuit,
Reference numeral 14 is an adder circuit, 15 and 16 are frame memories, and 17 is a switch.

The signal input from the image input terminal 1 is a digitalized image signal and is represented by, for example, 8 bits per pixel.

This image signal is supplied to the difference circuit 2 and the motion compensation circuit 5, and in the difference circuit 2, the first signal of the motion compensation circuit 5 is supplied.
Is subtracted from the motion-compensated prediction signal which is the output of

The output signal of the difference circuit 2, that is, the prediction error signal is processed by the valid / invalid block judgment circuit 3 in units of blocks of a predetermined size such as 8 lines × 8 pixels, for example, the average value of the prediction error absolute value within the block. Is calculated, and when this value is less than a predetermined threshold value, this block is determined to be an invalid block, the prediction error value in this block is zeroed, the other blocks are determined to be valid blocks, and the prediction error signal is used as it is. Is output to the encoding / quantization circuit 4.

The coding / quantization circuit 4 uses a predetermined coding method to which the present invention is applied, for example, discrete cosine transform (DC).
T) is performed, the encoded data is quantized by a predetermined characteristic, and the quantized output is output to the multiplexing circuit 6 and the dequantization / decoding circuit 13.

The multiplexing circuit 6 multiplexes the output of the coding / quantization circuit 4 and the motion vector information which is the second output of the motion compensation circuit 5.

The entropy coding circuit 7 executes run-length coding, which allocates a code corresponding to a continuous length of value zero, or a probability of occurrence depending on the probability of occurrence of a quantized output. A short code is assigned to high data and a long code is assigned to data with a low occurrence probability, and variable length code assignment is performed to shorten the average code length.

As described above, the data having a non-uniform code length is temporarily stored in the buffer memory 8 and sequentially read at a speed corresponding to the transmission speed.

The packetizer 9 outputs the output of the buffer memory 8 to the non-priority packet output terminal 10 in the non-priority block according to the priority / non-priority block control signal which is the third output of the motion compensation circuit 5. At other times, the data is output to the priority packet output terminal 11 and the interframe coded data is sent to the receiving side.

The dequantization / decoding circuit 13 receives the output signal of the coding / quantization circuit 4, and based on a predetermined method,
The reverse processing of the encoding / quantization circuit 4 is performed, and a prediction error signal including an error based on the above-mentioned invalid blocking and quantization is created.

This signal is added to the first output of the motion compensation circuit 5 in the addition circuit 14.

The output of the adder circuit 14, that is, the locally decoded signal is supplied to the frame memory 15 and the switch 17.

The switch 17 is connected to the output of the adder circuit 14 in the priority block, and is connected to the output of the frame memory 16 at other times.

Each of the frame memories 15 and 16 stores the input signal for one frame period and supplies it to the motion compensation circuit 5 as a prediction signal for the input signal of the next frame.

As is clear from the above description, the frame memory 15 stores the locally decoded signals for all blocks, whereas the frame memory 16 stores only the locally decoded signals for the priority block corresponding to a large motion. Will be done.

Next, one embodiment of the motion compensation circuit 5, which is a feature of the present invention, will be described.

FIG. 2 shows an embodiment of the motion compensation circuit 5, in which 51 is a motion vector detection circuit, 52 is a comparison circuit, and 5 is a comparison circuit.
Reference numeral 3 is a prediction signal selection circuit.

The motion vector detection circuit 51 uses the output image signal of the frame memory 15 supplied via the switch 54 for the input image signal input from the image input terminal 1 and, for example, a block of 16 lines × 16 pixels. The motion vector is detected for each.

That is, for the input block, the block having the smallest error among the neighboring blocks one frame before is detected as the optimum block, and the vector from this optimum block to the concerned block is used as the motion vector information for comparison circuit. 52, the prediction signal selection circuit 53, and the multiplexing circuit 6.

The comparison circuit 52 is a motion vector detection circuit 5
The motion vector information supplied from 1 is compared with a predetermined vector threshold value, and when the motion vector information is greater than or equal to the vector threshold value, the block is determined as a priority block, and when the motion vector information is less than the vector threshold value, it is determined as a non-priority block. It is determined and the priority / non-priority block control signal is output to the prediction signal selection circuit 53 and the packetizer 9.

The prediction signal selection circuit 53 selects the frame memory 15 from the outputs of the frame memories 15 and 16 when the priority / non-priority block control signal supplied from the comparison circuit 52 represents a non-priority block. , If it represents a priority block, select the frame memory 16,
Among the output signals of the selected frame memories, the output of the motion vector detection circuit 51, that is, the signal of the block corresponding to the vector designated by the motion vector information is selected, and this is used as the motion compensation prediction signal for the differential circuit 2 and Adder circuit 1
Output to 4.

When the refresh request signal is input from the refresh request signal input terminal 12, the predictive signal selection circuit 53 selects only the output of the frame memory 16.

At this time, the switch 54 is the frame memory 1
6 output is selected, otherwise, the frame memory 15
Select the output of.

In the above description, the case where the switch 54 normally connects the output of the frame memory 15 and therefore the motion vector detection circuit 51 detects the motion vector only for the output of the frame memory 15 has been described. , A motion vector is detected for both outputs of the frame memories 15 and 16, and when the detected vector represents the output of the frame memory 15, it is a non-priority block, and when it represents the output of the frame memory 16, Obviously, there is a method of determining a priority block.

Furthermore, in the above description, the case where discrete cosine transform is used in the coding / quantization circuit 4 has been described, but it is clear that other vector quantization method or previous value predictive coding method can be applied. is there.

[0040]

As described above, according to the present invention, the motion vector detected by the motion compensation circuit 5 is used to determine whether the relevant block being encoded is a priority block or a non-priority block. Therefore, there is no need to install a special circuit for the priority / non-priority block determination, and therefore, there is an advantage that the circuit configuration becomes simple.

[Brief description of drawings]

FIG. 1 is an example of the present invention.

FIG. 2 is an example of a motion compensation circuit.

[Explanation of symbols]

 1 image input terminal 2 difference circuit 3 valid / invalid block determination circuit 4 coding / quantization circuit 5 motion compensation circuit 6 multiplexing circuit 7 entropy coding circuit 8 buffer memory 9 packetizer 10 non-priority packet output terminal 11 priority packet output terminal 12 refresh request signal input terminal 13 inverse quantization / decoding circuit 14 addition circuit 15 frame memory 16 frame memory 17 switch 51 motion vector detection circuit 52 comparison circuit 53 prediction signal selection circuit 54 switch

Claims (1)

[Claims] 1. A frame memory for storing past images, the output of the frame memory is used to detect a motion vector of an input image signal to be encoded, and the detected motion is compensated. To form a prediction signal, encode a prediction error signal formed as a difference between the input image signal and the prediction signal by a predetermined method, and divide the encoded data into highly important components and low importance components. However, in a moving picture interframe coding apparatus that assigns a component of high importance to a priority cell and assigns a component of low importance to a non-priority cell, and transmits each block of a predetermined size with respect to the input image signal. A motion vector is detected, and the detected motion vector information and a motion compensation prediction signal are output, and the motion vector is compared with a predetermined vector threshold value, and is equal to or more than a threshold value. If the block is determined to be a priority block, the block is determined to be a non-priority block if less than the threshold value, and a motion compensation circuit that outputs a priority / non-priority block control signal indicating the determination result; A packetizer for allocating a signal of a priority block to a priority cell and a signal of a non-priority block to a non-priority block according to a priority block control signal, and a first decoded signal storing a locally decoded signal obtained by decoding prediction error signals of all blocks. A moving picture interframe coding apparatus comprising: a frame memory; and a second frame memory for storing a locally decoded signal obtained by decoding only the prediction error signal of the priority block.