JPH0746867B2 - Motion vector coding circuit - Google Patents

Description

DETAILED DESCRIPTION [Outline] In the present invention, motion detection is performed so that the code length can be sufficiently compressed even if the number of dropped frames in the motion vector encoder increases and the size of the motion vector increases. , A prediction error signal generating means, a quantized output coder, a motion vector coder, a multiplexing circuit and a buffer memory. And a plurality of encoders corresponding to the number of dropped frames when dropping frames, and all the encoders of the motion vector encoders are assigned to the motion vector of the size that occurs the most in each. It has a means for minimizing the code length, and is equipped with a counter and a switch for counting the number of dropped frames. The number of dropped frames by the counter is input to the multiplexing circuit, including the case where the dropped frames are not dropped. Together with is,
All of the encoders in the motion vector encoder are connected to the multiplexing circuit by the switch corresponding to the number of frames dropped, including the case where no frames are dropped. In the multiplexing circuit, the prediction error coded signal and the motion vector The coded signal and the number of dropped frames are multiplexed, and the multiplexed signal is sent to the transmission line via the buffer memory.

[Industrial application field]

The present invention multiplexes a signal obtained by encoding a prediction error in a frame or field unit of an image signal and a signal obtained by encoding a motion vector by a multiplexing circuit and inputs the signals into a buffer memory, and the buffer memory has a specified bit rate. The present invention relates to an improvement of a motion vector encoding circuit that drops frames until it can be transmitted at a specified bit rate if it is stored to the extent that it cannot be transmitted.

[Problems to be solved by conventional technology and invention]

FIG. 2 is a block diagram of a conventional motion vector encoding circuit, and FIG. 3 is a diagram showing the maximum range of motion vectors corresponding to the number of dropped frames.

In the figure, 1 is a motion detector, 2 is a variable delay device, 3 is a subtractor, 4
Is a quantizer, 5 is a frame memory, 6 is an adder, 7 is a quantized output encoder, 8 is a motion vector encoder, 9 is a multiplexing circuit, 10 is a buffer memory, 15 is a motion vector memory, SW1, SW2 , SW3 each represent a switch, and the variable delay device 2, the subtractor 3, the quantizer 4, the frame memory 5, and the adder 6 constitute a prediction error signal generating means.

In FIG. 2, the motion detector 1 detects the motion between the previous screen stored in the frame memory 5 and the current screen to be input, and this motion vector is passed through the motion vector memory 15 to the motion vector encoder 8 And the previous screen is delayed by this movement by the variable delay unit 2 to create a prediction screen, the difference between this screen and the current screen is obtained by the subtractor 3, and quantized by the quantizer 4, The error output is encoded and input to the multiplexing circuit 9.

The motion vector encoder 8 encodes the input motion vector, inputs it to the multiplexing circuit 9, and outputs it to the transmission path through the switch SW2 and the buffer memory 10 together with the encoded signal of the error output.

When a signal is stored in the buffer memory 10 to the extent that it cannot be transmitted at the specified bit rate, an encoding stop signal is sent from the buffer memory, the switch SW1 is turned to the 0 side, the switch SW2 is turned off, and frame dropping is performed. Also switch S
W3 is turned off while the signal stored in the buffer memory is sent to the transmission line.

Then, when the amount of stored signals decreases and the buffer memory 10 can transmit at the specified bit rate, the switches SW1 to SW3 are restored.

The motion detector 1 detects the motion with respect to the previous screen regardless of whether the frame is dropped or not, and sends out the detected motion vector to the motion vector memory 15. During the frame dropping period, the motion vector memory is already stored before the frame is dropped. The motion vector stored in 15 is used as the initial vector, and the initial vector and motion detector 1
The contents of the motion vector memory 15 are gradually updated so as to take the sum of the vectors sent from the other, and when no frames are dropped, the motion vector sent is entirely rewritten so as to be the initial vector. The motion vector is sent to the motion vector encoder 8. The motion vector encoder 8 encodes the code corresponding to the sent motion vector and inputs it to the multiplexing circuit, and at the same time, the error output is used for quantization output. It is sent to the transmission path together with the signal encoded by the encoder 7.

Since the size of the motion vector when not dropping frames is statistically small, the motion vector encoder 8 minimizes the code length when the magnitude of the motion vector is close to 0, but the number of dropping frames is small. As the number increases, QV1 is shown when the number of dropped pieces is 1, and QV when the number is 2 dropped.
2. When the number of dropped frames is n, the maximum range of the vector generated as a result of motion tracking like QVn increases, so the size of the motion vector increases statistically, so the code length becomes longer and Therefore, there is a problem that the code length is long and sufficient compression cannot be performed.

[Means for solving problems]

According to the present invention, the above problems are caused by a motion detector that detects a motion between the previous screen and the current input screen and outputs a motion vector, and a prediction screen generated by delaying the previous screen by a motion based on the output of the motion detector. Prediction error signal generating means for obtaining and quantizing a prediction error in a frame or field unit of the image signal from the difference between the current input screen and the current input screen, and the prediction error signal from the prediction error signal generating means is encoded and output as a prediction error encoded signal. Quantization output encoder, motion vector encoder that encodes the motion vector from the motion detector and outputs it as a motion vector encoded signal, multiplexing circuit to which the prediction error encoded signal and the motion vector encoded signal are input , And the output from the multiplexing circuit is stored and sent to the transmission line, but the amount of storage is larger than a predetermined threshold value and stored so that it cannot be transmitted at a specified bit rate. The output from the multiplexing circuit is cut off, frames are dropped in units of frames or fields, and the signal stored during that period is sent to the transmission line. In such a case, in a motion vector coding circuit having a buffer memory to which the output of the multiplexing circuit is input again, the motion vector encoder is divided into a coder in the case of not dropping frames and a frame dropping number in the case of dropping frames. Each of the motion vector encoders is composed of a plurality of corresponding encoders, and each encoder has a means for minimizing the code length assigned to the motion vector of the size that occurs most frequently. It is equipped with a counter and a switch that counts the number, and the number of dropped frames by the counter is input to the multiplexing circuit, including the case of not dropping frames, and all of the motion vector encoders Corresponding to the number of dropped frames, including the case where any of the encoders does not drop frames, the multiplexed circuit connects the prediction error coded signal, the motion vector coded signal, and the dropped frame number. This is achieved by a motion vector coding circuit characterized in that the multiplexed signal is sent to a transmission line through a buffer memory.

[Action]

According to the present invention, in the case of a motion vector having a magnitude corresponding to the number of frames or fields dropped, the encoder that minimizes the code length is set to the number of frames or fields dropped. (Including the case of not dropping pieces)
Corresponding to, the number of frames or fields dropped is detected, and the motion vector at this time is coded by the encoder corresponding to this number, so that it is transmitted in units of coded frames or fields. The code length can be shortened and the motion vector information can be compressed even if the number of frames or fields in which the number of frames is dropped is taken into consideration.

〔Example〕

FIG. 1 is a block diagram of a motion vector coding circuit according to an embodiment of the present invention.

In the figure, 9-1 is a multiplexing circuit, 11 to 13 are encoders in a motion vector encoder, 14 is a counter, SW4 is a switch, and the same reference numerals indicate the same functions throughout the drawings.

The points different from the case of FIG. 2 will be described with reference to FIG. 1 when the counter 14 that counts the number of frames dropped and the motion vector encoder, which does not drop frames, has a maximum motion vector value close to 0. , Which minimizes the code length of the encoder 11, the encoder 12 which minimizes the code length in the case of the motion vector having the largest size when the number of dropped frames is 1.
When the number of dropped frames is n, an encoder 13 is provided that minimizes the code length in the case of a motion vector of the size that occurs the most, and when not dropped, the switch SW4 selects the encoder 11, In the case of dropping, a multiplexer circuit corresponding to the number of dropped frames counted by the counter 14, for example, when the number of dropped frames is n, the encoder 13 is selected, and the motion vector is coded by each selected encoder. 9-1 and outputs to the other side together with the output of the quantized output encoder 7. At this time, the signal of the number of dropped frames counted by the counter 14 is also input to the multiplexing circuit 9-1 and transmitted to the partner side, and the partner side knows the characteristic of the motion vector encoded by this number. Decrypt.

Therefore, the code length of the motion vector is shortened, and the motion vector information can be compressed.

〔The invention's effect〕

As described in detail above, according to the present invention, it is possible to compress the motion vector information into the most efficient code length even when dropping frames.

[Brief description of drawings]

FIG. 1 is a block diagram of a motion vector encoding circuit according to an embodiment of the present invention, FIG. 2 is a block diagram of a conventional motion vector encoding circuit, and FIG. 3 is a maximum motion vector corresponding to the number of dropped frames. It is a figure which shows a range. In the figure, 1 is a motion detector, 2 is a variable delay device, 3 is a subtractor, 4 is a quantizer, 5 is a frame memory, 6 is an adder, 7 is a quantized output encoder, and 8 is a motion vector code. , 11, 12 and 13 are encoders in the motion vector encoder, 9 and 9-1 are multiplexing circuits, 10 is a buffer memory, 14 is a counter, 15 is a motion vector memory, and SW1 to SW4 are switches.

Claims (1)

[Claims] 1. A motion detector for detecting a motion between a previous screen and a current input screen and outputting a motion vector, a prediction screen generated by delaying the previous screen by a motion based on the output of the motion detector, and a current input. Prediction error signal generating means for obtaining and quantizing the prediction error of the image signal frame or field unit from the difference from the screen, Quantization output for encoding the prediction error signal from the prediction error signal generating means and outputting as the prediction error coded signal Encoder, a motion vector encoder that encodes the motion vector from the motion detector and outputs it as a motion vector encoded signal, a multiplexing circuit to which the prediction error encoded signal and the motion vector encoded signal are input, and multiplex The output from the multiplexing circuit is stored and sent to the transmission line, but if the amount of storage is larger than a predetermined threshold and is stored so that it cannot be transmitted at a specified bit rate, the multiplexing circuit Output is cut off, frames are dropped in units of frames or fields, and the signal stored during that period is sent to the transmission line. If the stored amount becomes less than a predetermined threshold value, the multiplexing circuit In a motion vector coding circuit having a buffer memory to which an output is input again, a motion vector coder is composed of a plurality of codes corresponding to the number of dropped frames and the number of dropped frames. Each of the motion vector encoders has a means for minimizing the code length assigned to the motion vector having the largest size in each, and a counter for counting the number of dropped frames. The number of frames dropped by the counter is input to the multiplexing circuit, including the case where frames are not dropped, and any of the encoders in the motion vector encoder is used. Is connected to the multiplexing circuit by a switch corresponding to the number of frames dropped, including the case where frames are not dropped. In the multiplexing circuit, the prediction error coded signal, the motion vector coded signal, and the number of dropped frames are about to be multiplexed. Motion vector coding circuit characterized in that the multiplexed signal is sent to a transmission line through a buffer memory.