JPH043593A - Method and apparatus for decoding moving picture - Google Patents

Abstract

PURPOSE:To reduce the number of components by using a two-dimension spatial filter as a post filter in a frame omission mode and using a two-dimensional filter as an in-loop filter in the decoding frame mode. CONSTITUTION:A switch 5 is closed in the decoding frame mode and a decoded data is written in a frame memory 8, and the switch 5 is opened in the low speed shot frame mode. Then movement is compensated in the frame omission mode and the resulting picture is outputted to a two-dimensional spatial filter 6 and a signal 108 from a compensation circuit 7 is converted into a signal 105 subjected to in-loop filter processing by the two-dimension spatial filter 6. Moreover, in the frame omission mode, no movement compensation is applied and the picture is outputted to the two-dimension spatial filter 6, output signals 109 of the two-dimensional spatial filter 6 are selected, post filter processing is applied to a reproduced picture 107 and the result is outputted to an output terminal 10 via a selection circuit 9. Thus, only one expensive two-dimensional spatial filter 6 is enough for the processing and the hardware is saved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] Ko Minoto relates to a video decoding method and apparatus used for video conferences and video telephones.

[Conventional technology]

2. Description of the Related Art In recent years, development of image transmission devices for the purpose of video conferences and video calls has been actively conducted.

In the encoding section of this image transmission device, a method is becoming standardized as an encoding algorithm in which a motion-compensated inter-frame difference signal is subjected to DCT transformation and then quantized. Furthermore, a two-dimensional spatial filter is used within the encoding loop to remove noise within the encoding loop caused by quantization. Furthermore, the decoding unit in the image transmission device applies a two-dimensional spatial filter to the reproduced image, in order to reproduce a visually good image.

FIG. 3 is a block diagram showing a conventional decoding method.

In the decoding system shown in FIG. 3, an encoded signal 200 input from an input terminal 11 is dequantized (Q-') in an inverse quantization circuit 12, and inversely DCT-transformed (DCT-') in an inverse DCT transformer circuit 13. '), the spatial domain difference signal 220
Play. Next, the reproduced difference signal 220 is sent to the adder 14
In, a two-dimensional in-loop filter (LFIL) 16
The reproduced image 23 is added with the signal 270 from
It becomes 0. This reproduced image 230 is stored in a frame memory (F
M) 18 and a post filter 30. The frame memory 18 stores the reproduced image of the previous frame. The reproduced image signal 250 of the previous frame from the frame memory 18 is motion compensated in the motion compensation circuit 17 and becomes a signal 260. This signal 260 is assumed to be a signal 270 filtered by the two-dimensional in-loop filter 16.

Here, the operation of the in-loop filter 16 is based on the motion vector (
MV) is controlled by an identification signal 280 indicating the presence/absence of the MV. This operates the in-loop filter 16 to perform in-loop filtering only when the MV identification signal 280 is on. Next, the reproduced image 230 is filtered by a two-dimensional post filter 30 and output. This post filter 30, like the in-loop filter, uses the MV identification signal 28
Controlled by 0.

[Problem to be solved by the invention]

In the above-mentioned decoding method, it is necessary to arrange two two-dimensional filters, the in-loop filter 16 and the post-filter 30, which have the same characteristics, and there is a problem in that the noise increases.

SUMMARY OF THE INVENTION An object of the present invention is to provide a video decoding method and device that eliminates the above drawbacks and reduces the number of parts.

[Means to solve the problem]

The video decoding method of the present invention obtains a reproduced image from a reproduced difference signal obtained from an input coded signal and a signal subjected to filter processing using an in-loop filter, and also performs post-filter processing on the reproduced image and outputs it. The moving image decoding method for obtaining a signal is characterized in that a two-dimensional spatial filter is used as a post-filter when dropping frames, and the two-dimensional filter is used as an in-loop filter when decoding frames.

The video decoding device of the present invention obtains a reproduced image from a reproduced difference signal obtained from an input coded signal and a signal subjected to filter processing using an in-loop filter, and also performs post-filter processing on the reproduced image and outputs it. A moving image decoding apparatus for obtaining a signal is characterized in that a two-dimensional spatial filter is used as a post filter when dropping frames, and the two-dimensional filter is used as an in-loop filter when decoding frames.

The video decoding device of the present invention includes an adder that obtains a reproduced image from a reproduced difference signal obtained from a manually input coded signal and a signal filtered by an in-loop filter, and an adder that converts the output from the adder into a frame frame. A switch that is turned on and output by a dropping signal, a frame memory that stores the reproduced image from this switch, a two-dimensional spatial filter that performs filter processing on a motion-compensated signal output from this frame memory, and It is characterized by comprising a selection circuit that outputs a signal from the frame memory and a signal from the two-dimensional filter at the time of decoding the frame.

The present invention focuses on two characteristics in the decoding method,
This filter doubles as an in-loop filter and a post filter. That is, firstly, an in-loop filter is not required for a dropped frame (a frame that is not decoded). Therefore, during this time, it becomes possible to assign a two-dimensional filter to the post filter.

Second, in image scenes with large movements, noise due to quantization increases, the number of dropped frames is large, and the noise removal effect of the in-loop filter is visually reduced; In small image scenes, there are few dropped frames, so the in-loop filter has a quantization noise removal effect.

Therefore, due to the above two characteristics, in the case of a scene with large motion, that is, a scene with large frame drop, the two-dimensional spatial filter 6 acts as a post filter during the frame drop, and the two-dimensional spatial filter 6 acts as an in-loop filter during the decoded frame. It acts as a. This results in
The same effect as the conventional example appears, and the decoding method can be realized with small hardware.

〔Example〕

Next, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a moving image decoding device of the present invention.

In FIG. 1, an input terminal 1 is connected to an inverse quantization circuit 2, and a code signal 101 inputted from the input terminal 1 is inversely quantized (Q-') L by the inverse quantization circuit 2, resulting in a signal 102. Output. The inverse quantization circuit 2 is connected to the inverse DCT transformation circuit 3, and the signal 10 from the inverse quantization circuit 2 is
2 is supplied to the inverse DCT conversion circuit 3. The signal 102 is subjected to inverse DCT transformation (DCT-') in the inverse DCT transformation circuit 3 and is made into a reproduced difference signal 103. The reproduced difference signal 103 is converted to the signal 105 from the two-dimensional spatial filter 6 in the adder 4.
are added to obtain the current frame reproduction signal 104.

The reproduced signal 104 is supplied to the frame memory 8 via the switch 5. Output signal 10 from frame memory 8
7 is supplied to a motion compensation circuit 7 and a selection circuit 9.

The signal 108 subjected to motion compensation by the motion compensation circuit 7 is filtered by the two-dimensional spatial filter 6 to form a signal 105, which is supplied to the adder 4 and the selection circuit 9.

The switch circuit 5 is controlled by a frame drop signal 111 supplied to the decoding circuit, and when a frame is to be decoded, the switch 5 is turned on, and decoded data is written to the frame memory 8. When a frame is to be dropped, the switch 5 is turned on and when a frame is to be dropped, the switch 5 is turned on. is turned off, and no decoded data is written to the frame memory 8. The selection circuit 9, like the switch circuit 5, is controlled by the frame frame dropping signal 111, and selects the output 107 from the frame memory 8 when the frame is to be decoded, and selects the output signal from the two-dimensional spatial filter 6 when the frame is being dropped. 109 and outputs the selection signal from the output terminal 10. Further, the two-dimensional spatial filter 6 will be explained in detail later, but
It smoothes the output signal 108 from the motion compensation circuit, and is controlled on/off by a motion vector (MV) identification signal 110, and is filtered when there is a motion vector, and is not filtered otherwise. It's being controlled. The motion compensation circuit 7 outputs a signal 108 in which motion is compensated for the signal 107 from the frame memory 8 based on the motion vector identification signal 110, and when the identification signal 110 is a decoded frame, it performs motion compensation and When the identification signal 110 is a dropped frame, it is output to the two-dimensional spatial filter 6 without motion compensation.

The operation of the decoding circuit configured in this way will be explained in detail.

The encoded signal 101 is passed through an inverse quantization circuit 2 and an inverse DCT transform circuit 3 to obtain a reproduced difference signal 103.

At the time of decoding the frame, the signal 108 is motion-compensated by the motion compensation circuit 7 with respect to the reproduced image 107 of the previous frame stored in the frame memory 8 as described above. The signal 108 from the motion compensation circuit 7 is a signal 105 that has been subjected to in-loop filter processing by the two-dimensional spatial filter 6.
becomes. This signal 105 and the reproduced difference signal 103 are added together by the adder 4, and thus the reproduced image 10 of the current frame is
4 is obtained. This reproduced image 104 of the current frame is stored again in the frame memory 8. Before this storage, the output of the frame memory 8 is outputted to the output terminal 10 via the selection circuit 9.

In the case of a dropped frame, as described above, the reproduced image 107 of the previous frame stored in the frame memory 8 is subjected to post-filter processing and outputted to the output terminal 10 via the selection circuit.

Here, details of the two-dimensional spatial filter 6 will be explained with reference to FIG. 2.

As described above, the two-dimensional spatial filter 6 is controlled on/off depending on the presence or absence of a motion vector. The motion vector is
The movement of the entire screen is detected as a small block of a certain size. Therefore, the filter is turned on/off for the corresponding small block. An example of the coefficients of the two-dimensional spatial filter 6 is as shown in FIG. A spatial filter can be realized by using the result of multiplication and accumulation as the pixel of interest.

The present invention can be implemented as described above.

〔Effect of the invention〕

As explained above, according to the present invention, the following effects are achieved.

In the video decoding method according to the first aspect, it is possible to reduce the amount of hardware.

In the moving image decoding device according to the second aspect of the invention, an expensive two-dimensional spatial filter can be omitted.

In the moving picture decoding apparatus according to the third aspect, an expensive two-dimensional spatial filter can be omitted, and decoding noise can be sufficiently removed.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an embodiment of the present invention, FIG. 2 is a block diagram explaining the contents of a two-dimensional spatial filter used in the present invention, and FIG. 3 is a block diagram showing a conventional device. 1... Input terminal, 2... Inverse quantization circuit, 3... Inverse DCT transform circuit, 4... Addition circuit, 5... Switch, 6... Two-dimensional spatial filter, 7... Movement Compensation circuit, 8... Frame memory, 9... Selection circuit, 10... Output terminal.

Claims (1)

[Claims] 1. A reproduced image is obtained from a reproduced difference signal obtained from an input code signal and a signal subjected to filter processing by an in-loop filter, and the reproduced image is subjected to post-filter processing to output a signal. A video decoding method for obtaining the following: a two-dimensional spatial filter is used as a post-filter when dropping frames, and the two-dimensional filter is used as an in-loop filter when decoding frames. 2. A video decoding device that obtains a reproduced image from a reproduced difference signal obtained from an input code signal and a signal subjected to filter processing using an in-loop filter, and also performs post-filter processing on the reproduced image to obtain an output signal. In , a two-dimensional spatial filter is used as a post filter at the time of frame dropping,
A video decoding device characterized in that the two-dimensional filter is used as an in-loop filter when decoding frames. 3. An adder that obtains a reproduced image from the reproduced difference signal obtained from the input code signal and the signal filtered by the in-loop filter, and outputs the output from the adder as turned on by a frame frame dropping signal. A switch, a frame memory that stores the reproduced image from this switch, a two-dimensional spatial filter that performs filter processing on a motion-compensated signal output from this frame memory, and a decoded frame memory that decodes the signal from the frame memory at the time of a frame drop-off frame. 1. A moving image decoding device comprising: a selection circuit that outputs a signal from the two-dimensional filter at the time of a frame.