JPS63294086A - Interframe encoding system - Google Patents

Abstract

PURPOSE:To delete the quantity of transmission and to attain an efficiency by using the output index of a vector quantizer as the selective index of an arthogonal transforming function. CONSTITUTION:When a predictive error signal between frames is inputted to the vector quantizer 5 and an orthogonal transformer 1, the output signal of the orthogonal transformer 1 is selected in a selector 6 according to the output signal of the vector quantizer 5. Thereafter, it is transmitted to a quantizer 2, an encoder 3. The other output of the quantizer 2 is fed to a predictive device 4 and reversely quantized, to reversely execute the orthogonal transformation of the orthogonal transformer 1 selected according to the output index of the vector quantizer 5 to perform a frame reproduction and prepare for a next encoding between the frames.

Description

[Detailed description of the invention] [Summary] An interframe prediction error signal of an input signal such as an image signal converted into a codigital signal is transformed by a plurality of orthogonal transformers, and vector quantization is performed corresponding to each orthogonal transformer. The orthogonal transform output is selected in accordance with the output index of the converter, and encoded and transmitted, and only the output index is encoded and transmitted depending on the situation.

[Industrial Application Field] The present invention relates to an interframe coding method, and particularly to a function adaptive interframe coding method for highly efficient coding of an interframe prediction error signal.

[Prior Art] A conventional method for performing interframe coding is shown in FIG. 5, in which 51 is an orthogonal transformer, 52 is a quantizer,
53 is a subtracter for calculating the error with the predicted signal, 54 is an encoder, 55 is a buffer memory, 56 is an inverse quantizer, 57 is an inverse orthogonal transformer, 58 is an adder, 59 is a frame memory,
60 is a variable delay circuit, and 61 is a motion amount detector.

In operation, an input signal such as a video signal converted into a digital signal is applied to the subtracter 53 and the motion amount detector 61. The motion amount detector 61 detects the amount of motion of the current frame with respect to the video signal of the previous frame stored in the frame memory 59 (actually, a variable delayed video signal to be described later), and uses this motion amount information to detect the amount of motion of the current frame. 60 is controlled. The video signal of the previous frame read from the frame memory S9 is applied as a motion-compensated prediction signal to the subtracter 53 via the variable delay circuit 60, and subtracted from the input signal. An interframe prediction error signal is output from the subtracter 53 and applied to the orthogonal transformer 51.

The interframe prediction error signal subjected to orthogonal transformation by the orthogonal transformer 51 is applied to a quantizer 52 to be quantized, and then applied to an encoder 54 and an inverse quantizer 56. encoder 5
4, a coded signal obtained by coding the quantized output signal from the quantizer 52 and a coded signal obtained by coding the motion amount detection information from the motion amount detector 61 are time-division multiplexed and sent to the buffer memory 55. It is sent to the transmission line via this buffer memory 55 at a predetermined transmission speed.

Further, the quantized output signal from the quantizer 52 is subjected to a quantization process inverse to the above quantization by an inverse quantizer 56, and is further subjected to a lotus orthogonal transformation process by an inverse orthogonal transformer 57. After that, the adder 58 adds the variable delay circuit 6
The signal is added to the video signal of the previous frame output via 0 and stored in the frame memory 59.

Note that the inverse quantizer 56, the inverse orthogonal transformer 57, the frame memory 59, the variable delay circuit 60, and the motion amount detector 61 constitute a predictor that performs playback prediction on the receiving side.

In addition, as in the above configuration example, by introducing motion amount detection into interframe coding, band compression can be performed even in video signals with large motion.
It goes without saying that inter-frame encoding can be performed without using the variable delay circuit 60.

[Problems to be Solved by the Invention] As mentioned above, by orthogonally transforming and quantizing the interframe prediction error signal and, if necessary, compensating for the amount of motion of the video signal of the current frame by detecting the amount of motion. , it is possible to encode and transmit a video signal with high efficiency and perform faithful image reproduction on the receiving side, but it is not efficient to encode all interframe prediction error signals with a single orthogonal transformer. It doesn't necessarily mean it's good. This is because the optimal orthogonal transform function for each block obtained by subdividing the interframe prediction error signal is not necessarily the same as the orthogonal transform function used in the orthogonal transformer.

Therefore, one object of the present invention is to realize a system for efficiently encoding and transmitting a prediction error signal that cannot be compensated for by a single orthogonal transformer.

[Means for Solving the Problems] In order to achieve the above object, the first interframe coding method of the present invention uses orthogonal A transform device 1, a vector quantizer 5 which inputs an inter-frame prediction error signal at the same time as the orthogonal transform device 1, and performs quantization using a code book having an output index corresponding to each orthogonal transform function, and this vector quantizer. A selector 6 that selects one of the orthogonal transformer groups according to the output index of 5,
The predictor 4 is also configured to perform an orthogonal transformation inverse to the selected orthogonal transformer.

FIG. 2 conceptually shows an interframe coding method according to the second invention. In addition to the first invention, in this second invention, the quantization output of the vector quantizer 5 is A vector inverse quantizer 7 that performs inverse quantization, a subtracter 8 that generates a residual signal between the output of the inverse quantizer 7 and the interframe prediction error signal, and a vector quantizer in accordance with the residual signal. The encoder 3 is controlled to select the output index of the vector inverse quantizer 7 or the output of the quantizer 2, and the predictor 4 also selects the output of the vector inverse quantizer 7 or the inverse orthogonal transform corresponding to the output index. and a controller for controlling the selection of the output.

[Operation] In the first interframe coding method of the present invention, in FIG. 1, an interframe prediction error signal is input to an orthogonal transformer 711 comprising a vector quantizer 5 and a group of orthogonal transformers. This group of orthogonal transformers each has a −L transform function for which the output index of the vector quantization code is known to be the same for the interframe prediction error signal, so the vector quantizer There is a one-to-one correspondence between each output index of the codebook No. 5 and the transformer to be selected in the orthogonal transformer group. Therefore, the selector 3 selects the output signal of the orthogonal transform device 1 according to the output signal of the vector quantizer 5. After that, quantizer 2 and encoder 3
transmitted to. The other output of the quantizer 2 is sent to the predictor 4, where it is inversely quantized and subjected to inverse orthogonal transformation by the orthogonal transformer selected at the top according to the output index of the vector quantizer 5. Performs playback and prepares for the next interframe encoding.

In the interframe coding method according to the second invention shown in FIG.
When the vector code is reproduced by the vector inverse quantizer 7 from the output index when quantized by A control signal for selecting the output index of vector quantization or the output of the quantizer 2 is given to the encoder 3 in accordance with the output index of the vector quantization, and encoding is performed for sending out to the transmission path. This is also performed in the reproduction operation in the predictor 4, and the predictor 4 performs an encoding operation on the output of the inverse orthogonal transformer based on the vector quantization index or the residual signal due to the vector inverse quantization according to the control signal from the controller 9. Similarly, frames are selected and reproduced in preparation for the next interframe encoding.

[Implementation] Hereinafter, embodiments of the interframe coding method according to the present invention will be described in detail with reference to the drawings.

FIG. 3 is a block diagram showing an embodiment of the first invention, in which the same symbols as in FIG. 1 indicate the same parts, and the orthogonal transformer 1 in FIG. The predictor 4 includes an inverse quantizer 20 that performs inverse quantization of the quantizer 2, and n inverse orthogonal transformers that perform inverse orthogonal transform corresponding to each orthogonal transformer of the orthogonal transform device 1. Inverse orthogonal transform device 2 consisting of a group of transformers
1, a selector 22 that adaptively selects one inverse orthogonal transformer from a group of inverse orthogonal transformers according to the output index of the vector quantizer 5, and an adder 23 similar to the conventional example,
It includes a frame memory 24, a variable delay circuit 25, and a motion amount detector 26.

Next, to explain the operation, the input signal converted into a digital signal is applied to the subtracter 11 and the motion amount detector 26, and the predicted signal is motion-compensated using the motion amount detector 26 and the variable delay circuit 25. An interframe prediction error signal of the difference between and the input signal is output from the subtracter 11 and sent to the preprocessor
added to.

The preprocessor 10 is a noise removal converter that removes noise components by outputting, for example, an interframe prediction error signal whose absolute value is minute as O, and is provided when deemed necessary.

The output of the preprocessor 10 is input to each orthogonal transformer of the orthogonal transform device 1 and the vector quantizer 5. The vector quantizer 5 extracts (matches) the vector code that best matches the input prediction error signal from a codebook (not shown).
The index is used as the output signal. This output signal is input to the selector 6 and the selector 22 of the predictor 4.

The selector 6 uses the output signal of the vector quantizer 5 to select n
One of the orthogonal transform output signals is adaptively selected and sent to the quantizer 2 and encoder 3. Here, each orthogonal transformer in the orthogonal transform device 1 is configured with, for example, a −L transform function for which the output index of the vector quantizer 5 is known to be the same for the interframe prediction error signal. Therefore, by using the output index for each vector code in the codebook of the vector quantizer 5, it is possible to uniquely select the transformation function that best matches the prediction error signal that is currently being encoded.

The output signal of the quantizer 2 is input to the encoder 3 as well as to the inverse quantizer 20 . Each output of the inverse quantizer 20 is input to an inverse orthogonal transform device 21 , each of which undergoes inverse orthogonal transform and is input to a selector 22 .

The selector 22 sends the inverse orthogonal transformed signal selected by the output index from the vector quantizer 5 to the adder 23 to reproduce the previous frame screen in the same manner as in the conventional example.

In the encoder 20, the output of the quantizer 2 and the motion amount detector 2
The signals from 6 are encoded and multiplexed to the buffer memory (
(not shown) and sends it out to the transmission path.

In this way, by using the output index of the vector quantizer as an index for selecting orthogonal transform functions, various orthogonal transform functions can be adaptively controlled.

FIG. 4 is a block diagram showing an embodiment of the interframe coding method according to the second invention, and the predictor 4 shown in FIG. an inverse orthogonal transform device 21 consisting of n inverse orthogonal transformers that perform inverse orthogonal transform corresponding to each orthogonal transformer of the orthogonal transform device 1; a selector 22a that inputs the output index of , the vector code reproduced by the vector dequantizer 7, and control information from the controller 9, an adder 23, a frame memory 24, and a variable delay circuit 25 similar to the conventional example. , and motion amount detector 26
It consists of

In operation, similarly to the first invention, the orthogonal transformer output adaptively selected by the selector 6 according to the output vector code index of the vector quantizer 5 is quantized by the quantizer 2 and then sent to the encoder. Sent to 3.

On the other hand, the output index of the vector quantizer 5 is sent to the vector dequantizer 7, the vector code used in the vector quantizer 5 is reproduced, and the residual signal between this and the interframe prediction error signal is subtracted. It is obtained by the controller 8 and inputted to the controller 9.

The controller 9 calculates the power of the residual signal, and this is the predetermined threshold T! If it is larger, the quantized output from the quantizer 2 is sent to the transmission path.

On the other hand, if the power of the residual signal is smaller than the threshold T1, the control ilS controls the encoder 3 so as to send only the output index of the vector quantizer 5 to the transmission path.

In order to execute the above encoding operation also in the reproduction operation in the predictor 4, control information is sent from the controller 9 to the selector 22a. This control information selects the inverse orthogonal transformer output according to the output index of the vector quantizer 35, or this inverse orthogonal transform operation is stopped and the vector inverse quantizer 7
It contains information on whether to select the vector code output from the vector code, and is executed in correspondence with the encoding operation.

In this way, when the residual signal from vector quantization is small, the index of the vector code with less information is transmitted without transmitting the orthogonal transform output, making interframe coding by orthogonal transform more efficient. making it good.

Effect of the Invention 1 As explained above, in the present invention, by using the output index of the vector quantizer as a selective index of the orthogonal transform function, the optimal one is selected from among many orthogonal transform functions. Furthermore, when the residual signal during vector quantization is small, only the vector quantization output index is transmitted without using orthogonal transformation, which reduces the amount of transmission and creates a more efficient code. method can be realized.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the concept of the interframe coding method according to the first invention, FIG. 2 is a block diagram showing the concept of the interframe coding method according to the second invention, and FIG. is a block diagram showing an embodiment of the first invention, FIG. 4 is a block diagram showing an embodiment of the second invention, and FIG. 5 is a block diagram showing an example of the configuration of a conventional interframe coding method. Figure. In FIGS. 1 and 2, 1 is an orthogonal transform device, 2 is a quantizer, 3 is an encoder, 4 is a predictor, 5 is a vector quantizer, 6 is a selector, and 7 is a vector dequantizer. vessel, 8 is reduced r
tvs and 9 indicate a controller, respectively. In the drawings, the same reference numerals indicate the same or corresponding parts.

Claims (5)

[Claims] (1) Input signal converted to digital signal and predictor (
An interframe code that orthogonally transforms the interframe prediction error signal with the reproduced signal from 4) by an orthogonal transformer (1), quantizes it by a quantizer (2), and encodes and transmits it by an encoder (3). In the orthogonal transform system, the orthogonal transform device (1) includes a plurality of orthogonal transformers, and the prediction error signal is orthogonally transformed by each orthogonal transformer, and at the same time a vector quantizer (5
), and the output of each orthogonal transformer is vector quantized in a selector (6) according to the output index of the vector code of the vector quantizer (5) corresponding to each orthogonal transform function.
) is selected and sent to the quantizer (2), and the predictor (4) also performs inverse quantization and inverse orthogonal transformation according to the output index. . (2) The interframe coding method according to claim 1, wherein the predictor (4) detects the amount of motion of a frame and compensates the frame position by the amount of motion. (3) Input signal converted to digital signal and predictor (
The interframe prediction error signal with the reproduced signal from 4) is orthogonally transformed in the orthogonal transform device (1) consisting of a plurality of orthogonal transformers, and at the same time vector quantized in the vector quantizer (5). The output is selected by a selector (6) according to the output index of the vector code of the vector quantizer (5) corresponding to each orthogonal transform function, quantized by the quantizer (2), and then sent to the encoder (3). ), a vector dequantizer (7) reproduces a vector code from the output index of the vector quantizer (5) and generates a residual signal from the prediction error signal. The controller (9) converts the output index or the output of the quantizer (2) into the encoder (2) according to the residual signal generated by the subtracter (8).
3) controls the selection of the predictor (4).
Also in the interframe coding, the output of the vector inverse quantizer (7) or the output by inverse orthogonal transformation according to the output index is selected depending on the control output of the controller (9). method. (4) The interframe control according to claim 3, wherein the controller (9) determines the power of the residual signal and generates a control output depending on the magnitude of the power of the residual signal and a predetermined threshold. Encoding method. (5) The interframe coding method according to claim 3, wherein the predictor (4) detects the amount of motion of a frame and compensates the frame position by the amount of motion.