JPS63199589A - Interframe coding system - Google Patents

Abstract

PURPOSE:To improve coding efficiency and to eliminate the deterioration in the visual resolution of a reproduced picture by means of a filter by providing the filter in the front stage of a prediction device in a decoding loop, adaptively controlling a selector selecting a signal on an input side and that on an output side, and outputting a repro duced signal. CONSTITUTION:The filter 1 is provided in the front stage of the prediction device 2 in the decoding loop adding the signal from a coding or decoding part 5 and a prediction value to an adder 4 and executing decoding. The selector 3 which executes selection and output between the input side or the output side by a control signal is provided, and the selector 3 is adaptively controlled, whereby either of the signal on the input side and that on the output side of the filter 1 is selected so as to be set as the reproduced signal. Namely, since the detail of the received and reproduced picture cannot visually be recognized in the case of the picture with a big movement, the signal which the filter 1 has processed is selected by the selector 3 and is set to be the reproduced signal, and it is desirable to visually recognize the detail of the received and reproduced signal in the case of the picture with less movement, whereby the signal which the filter 1 does not process is selected and set to be the reproduced signal by the selector 3.

Description

[Detailed Description of the Invention] [Summary] In a method in which a filter is provided before a predictor in a decoding loop to perform interframe encoding or decoding, signals before and after filter processing are converted into motion vectors. A reproduction signal is adaptively selected based on information and the like, and for example, in a state where the resolution of a reproduction screen is reduced due to filter processing, the signal before filter processing is selectively output.

[Industrial application field]

The present invention relates to an interframe coding method that provides a filter before a predictor and performs coding or decoding between four seams.

In interframe coding methods that improve coding efficiency by reducing impulse-like noise caused by prediction by providing an in-loop filter, depending on the type of image, the resolution of the playback screen may drop noticeably. . Therefore, it is desired to prevent such a decrease in resolution.

[Conventional technology]

In an interframe encoding method that performs encoding or decoding between frames, a method is adopted in which an in-loop filter is provided to reduce impulse noise and improve encoding efficiency. FIG. 6 is a block diagram of the main part of the conventional example, showing the transmitting side of the conventional example. In the figure, 51 is a filter, 52 is a predictor including a frame memory, 53 is a subtracter, 54 is an adder, 55 is an inverse quantizer, 56 is a quantizer, 57 is an encoder,
A configuration in which a filter 51 is provided before a predictor 52 is shown.

In such a configuration, when an input signal from a television camera or the like is applied to the subtracter 53, an inter-frame difference signal from the predicted signal from the predictor 52 is obtained, and this difference signal is sent to the quantizer. 56, and the quantized output signal is encoded by an encoder 57 to become an output signal to be sent to the receiving side, and is dequantized by an inverse quantizer 55 to reproduce an interframe difference signal. . This difference signal is added to the predicted signal in an adder 54, becomes a locally decoded signal, and is applied to a filter 51, filtered, and applied to a predictor 52, as well as a reproduction signal applied to a monitor, etc. It becomes a signal.

Furthermore, when transmitting a moving image, the interframe difference signal becomes large, so a block matching type motion compensated interframe coding method is adopted. This method calculates the motion vector of a block of a predetermined size and performs interframe prediction by shifting the position of the predicted value according to the motion vector, so that prediction efficiency does not decrease even in the case of images with large motion. This is what I did. Also, along with the encoded signal of the difference signal,
It sends motion vector information to the receiving side.

On the receiving side, an inter-frame difference signal can be obtained by decoding and dequantizing the received encoded signal, and by adding this inter-frame difference signal and the signal from the frame memory. The configuration is such that the addition output is filtered using a filter and added to the frame memory, and the configurations of the inverse quantizer 55, adder 54, filter 51, and predictor 52 are the same on the receiving side, and the reproduced signal is the filtered signal. was used. Furthermore, when a block matching type motion compensated interframe coding method is used, a variable delay device for performing motion compensation is added on the receiving side.

[Problem that the invention seeks to solve]

As mentioned above, in the interframe coding method in which the filter 51 is provided in the local decoding loop, the output signal of the filter 51 is used as the reproduced signal, and this filter 5
1 removes impulse-like noise, but in the case of still images or similar images with little movement, the processing by the filter 51 suppresses high frequency components, resulting in a reduction in resolution.

An object of the present invention is to perform filter processing without reducing visual resolution.

[Means for solving problems]

The interframe encoding method of the present invention will be explained with reference to FIG. 1. The interframe encoding method of the present invention will be described with reference to FIG. A filter 1 is provided at the front stage of the filter 2, and a selector 3 is provided which selects and outputs between the input side and the output side of the filter 1 using a control signal.The selector 3 is adaptively controlled. Either the input side signal or the output side signal is selected and used as a reproduced signal.

[Effect]

In the case of an image with a large amount of movement, the details of the received and reproduced image cannot be visually recognized, so the signal processed by the filter 1 is selected by the selector 3 as the reproduced signal, and in the case of an image with little movement, the details of the received and reproduced image are Since it is desirable to be able to visually recognize the details of an image, the selector 3 selects the signal that is not processed by the filter 1 and uses it as a reproduced signal.The filter processing improves the encoding efficiency without reducing the visual resolution. can be improved.

〔Example〕

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

FIG. 2 is a block diagram of the main parts of the transmitting side of an embodiment of the present invention,
FIG. 3 is a block diagram of the main part of the receiving side, and shows a case where a block matching type motion compensated interframe coding method is applied. In each figure, 10 is a subtracter, 11
．． 21 is a filter, 12.22 is a frame memory, 13
．． 23 is a selector, 14.24 is an adder, 15.25 is an inverse quantizer, 16 is a quantizer, 18.28 is a determiner, 19
．． 27 is a variable delay device, 20 is a motion compensator, 17 is an encoder, and 26 is a decoder.

The filter 11.21 is the frame memory 12.2 of the predictor.
A selector 13.23 is provided before the filter 2 and selects the input side signal and the output side signal of the filter 113, 21.
is controlled by a control signal from the determiner 18.28,
The selected signal is used as a reproduced signal.

On the transmitting side, an input signal from a television camera or the like is applied to a subtracter 10 and a motion compensator 20. The motion compensator 20 matches the input signal and the signal from the frame memory 12, for example, in a block of 8×8 pixels, and
A motion vector for this block is determined, and the variable delay device 19 is controlled to correct the signal from the interframe difference signal 2 according to the motion vector, and the result is used as a predicted value to be added to the subtracter 10 and the adder 14. Furthermore, the motion vector information from the motion compensator 2o is added to the determiner 18 to determine whether or not the motion is large. Furthermore, the motion vector information is added to the encoder 17 via a path not shown and is sent to the receiving side.

The subtracter 10 outputs an inter-frame difference signal between the input signal and the motion-compensated predicted value, which is quantized by the quantizer 16, and the quantized output signal is encoded by the encoder 17 and sent to the receiving side. as well as an output signal to the inverse quantizer 15.
The signal is inversely quantized by
4 is added to the predicted signal to obtain a locally decoded signal.

This locally decoded signal becomes the input signal of the filter 11,
For example, spatial filter processing is performed in which the weight of the pixel of interest is set to 4, the weight of the four surrounding pixels is set to 1, and the arithmetic mean value of the brightness of the five pixels is set as the brightness of the pixel of interest. This filtered output signal is added to the frame memory 12 and held for one frame.

The locally decoded signal on the input side of the filter 11 and the filtered output signal on the output side are added to the selector 13, and the motion vector information from the motion compensator 20 is added to the determiner 18 to determine the block correspondence. It is determined whether the signal has a large movement. When the movement is large, the output signal of the filter 11 is selected and becomes the reproduction signal, and when the movement is small, the signal on the input side of the filter 110 is selected and becomes the reproduction signal.

On the receiving side, the received signal is applied to a decoder 26 to be decoded, and dequantized by an inverse quantizer 25 to become an interframe difference signal, which is applied to an adder 24. Further, motion vector information from the transmitting side is decoded by a decoder 26 and added to a variable delay unit 27 and a determiner 28.

The variable delay 2S27 is for delaying the signal of one bitroque read from the frame memory 22 according to the motion vector information and outputting a motion-compensated predicted value. The adder 24 adds the inter-frame difference signals, and the added output signal becomes a decoded signal. This decoded signal is filtered by a filter 21 and added to a frame memory 22, where it is held for one frame.

The determiner 28 determines whether the image has a large motion or a small motion based on the motion vector information and applies a control signal to the selector 23, and the control signal 43 causes the selector 23 to The output signal of the filter 21 is selectively outputted as a reproduced signal, and the signal on the input side of the filter 21 is selectively outputted as a reproduced signal when the movement is small, so that control is performed in a block-by-block manner.

Therefore, on the transmitting side and the receiving side, based on the motion vector information, the input side signal and the output side signal of the filter 11.21 are adaptively selected by the selector 13.23 to become a reproduced signal, For images with large movements, the changes cannot be visually recognized, so the filtered signal is used as the reproduction signal, and for images with small movements, even minute parts can be visually recognized, so the signal before filtering is used for reproduction. The encoding efficiency can be improved by filtering the signal without reducing the visual resolution.

Furthermore, this embodiment utilizes motion vector information in a block matching motion compensated interframe coding method, and this motion vector information is transmitted from the transmitting side to the receiving side together with the interframe coded signal. Therefore, there is no need to transmit a special signal to control the selector 23 from the transmitting side to the receiving side.

FIG. 4 is a block diagram of the main parts on the transmitting side of another embodiment of the present invention, and FIG. 5 is a block diagram of the main parts on the receiving side, where 30 is a subtracter, 31.41 is a filter, and 32.42 is a frame. Memory, 33.43 selector, 34.44 adder, 35.
45 is an inverse quantizer, 36 is a quantizer, 37 is an encoder,
38 is a determiner, 39.40 is a subtracter, 46 is a decoder,
47 is a determination controller.

This embodiment is for the case where no motion compensation is performed, and the operations of encoding and decoding between frames are similar to those of the previous embodiment. In this embodiment, the difference between the input side signal and the input signal of the filter 31 and the difference between the output side signal and the input signal are obtained in subtracters 39 and 40, respectively. The output signals di and d2 of 40 are added to the determiner 38, and when di<d2, the signal on the input side of the filter 31 is selected, and when d1≧d2, the signal on the input side of the filter 31 is selected.
A control signal is applied from the determiner 38 to the selector 33 so as to select the signal on the output side.

In other words, the signal with the smaller difference from the input signal is selected, and if the movement is small, the filter 31 suppresses the high frequency component, so the signal d1, which is the difference between the input signal and the locally decoded signal, is selected. , the difference signal d2 between the input signal and the output signal of the filter 31 becomes larger. Therefore, in this case, the signal on the input side of the filter 31 is selectively output as a reproduced signal. When the motion is large, the difference signal d1 between the input signal and the locally decoded signal becomes larger than the difference signal d2 between the input signal and the output signal of the filter 31. Therefore, the signal on the output side of the filter 31 is selectively output as a reproduced signal.

The control signal applied to the selector 33 is transmitted to the receiving side via the encoder 37, similar to the motion vector information described above.

On the receiving side, the signal is decoded by a decoder 46, dequantized by a dequantizer 45, and added to the predicted value from the frame memory 42 by an adder 44 for decoding. In addition, in the decoder 46, the control signal from the transmitting side is separated and applied to the judgment controller 47, which controls the selector 43 to determine whether the signal is on the input side or the output side of the filter 41. The signal is selectively output as a reproduction signal.

〔Effect of the invention〕

As explained above, the present invention provides a filter 1 before the predictor 2 in the decoding loop, and a selector 3 that selects the input side signal and the output side signal of the filter 1. The selector 3 is adaptively controlled to output a reproduced signal, and the filter 1 improves the encoding efficiency, and the signals before and after filter processing are adaptively selected as the reproduced signal. This has the advantage of not reducing the visual resolution of the reproduced image. Further, adaptive control of the selector 3 can be easily performed using motion vector information, a difference signal between an input signal and a signal before and after filter processing, or the like.

[Brief explanation of the drawing]

FIG. 1 is a diagram illustrating the principle of the present invention, FIG. 2 is a block diagram of essential parts on the transmitting side of an embodiment of the present invention, and FIG. 3 is a block diagram of essential parts on the receiving side of an embodiment of the present invention. FIG. 4 is a block diagram of the main parts of the transmitting side of another embodiment of the present invention, FIG. 5 is a block diagram of the main parts of the receiving side of another embodiment of the invention, and FIG. 6 is a block diagram of the main parts of the conventional example. It is. 1.11, 21, 31.41 are filters, 2 is a predictor,
12, 22, 32.42 is frame memory, 3, 13,
23, 33.43 is a selector, 4゜14.24.34.4
4 is an adder, 5 is an encoding or decoding unit, 15.25, 3
5.45 is an inverse quantizer, 16.36 is a quantizer, 17.
37 is an encoder.

Claims (3)

[Claims] (1) In the interframe encoding method in which a filter (1) is provided before a predictor (2) equipped with a frame memory in a decoding loop to perform interframe encoding or decoding, the above-mentioned A selector (3) is provided which selects and outputs signals on the input side and output side of the filter (1) using a control signal, and the selector (3) is adaptively controlled to output signals from the selector (3). An interframe encoding method characterized by outputting a reproduced signal. (2) The interframe encoding method according to claim 1, wherein the selector (3) is controlled in accordance with motion vector information. (3) The selector (3) is controlled by difference information between the input signal and the input and output signals of the filter (1), respectively. The interframe encoding method according to item 1.