Encoding method, decoding method, encoder, decoder and storage medium

Patent metadata

Field	Value
Publication number	US-12452434-B2
Application number	US-202318141098-A
Country	US
Kind code	B2
Filing date	Apr 28, 2023
Priority date	Oct 30, 2020
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

A decoder decodes a bitstream of a video and determines decoding parameters of a current picture in the video; determines, according to the decoding parameters of the current picture, reconstructed data of a second picture component of the current picture based on the reconstructed data of the first picture component by using a prediction model; determines decoded data based on the reconstructed data of the first picture component and the reconstructed data of the second picture component. An encoder determines identification information of a current picture in a video; codes reconstructed data of a first picture component; the encoder skips encoding of the reconstructed data of the second picture component, in response to the identification information indicating using the prediction model for determining the reconstructed data of the second picture component; and signals coded bits into a bitstream of the video.

First claim

The invention claimed is:

1. A decoding method, applied to a decoder, the method comprising:
decoding a bitstream of a video and determining decoding parameters of a current picture in the video, wherein the decoding parameters of the current picture comprises reconstructed data of a first picture component of one or more pictures in the video;
determining, according to the decoding parameters of the current picture, reconstructed data of a second picture component of the current picture based on the reconstructed data of the first picture component by using a prediction model, wherein the second picture component differs from the first picture component; and
determining decoded data of the current picture based on the reconstructed data of the first picture component and the reconstructed data of the second picture component of the current picture,
wherein the decoding parameters of the current picture further comprise identification information, wherein the identification information indicates whether to use the prediction model for determining the reconstructed data of the second picture component.

2. The method of claim 1 , wherein the first picture component is a luma component and the second picture component is a chroma component, or,
the first picture component is the chroma component and the second picture component is the luma component,
wherein the first picture component is a first colour component, the second picture component is a second colour component, and the second colour component differs from the first colour component.

3. The method of claim 1 , wherein decoding the bitstream of the video and determining the decoding parameters of the current picture in the video comprises:
decoding the bitstream of the video and obtaining the identification information from a data unit corresponding to the current picture.

4. The method of claim 1 , wherein decoding the bitstream of the video and determining the decoding parameters of the current picture in the video comprises:
decoding the bitstream of the video and determining a time layer attribute of the current picture, wherein the time layer attribute indicates whether the current picture is a high time layer picture or a low time layer picture; and
determining the identification information according to the time layer attribute.

5. The method of claim 4 , wherein determining the identification information according to the time layer attribute comprises:
setting the identification information to indicate using the prediction model for determining the reconstructed data of the second picture component, in response to the reconstructed data of the second picture component of the current picture not being included in reconstructed data of the current picture obtained by decoding the bitstream of the video, and the time layer attribute indicating that the current picture is the high time layer picture.

6. The method of claim 1 , wherein determining, according to the decoding parameters of the current picture, reconstructed data of the second picture component of the current picture based on the reconstructed data of the first picture component by using the prediction model comprises:
in response to the identification information being set to indicate using the prediction model for determining the reconstructed data of the second picture component, inputting the reconstructed data of the first picture component and reconstructed data of one or more low time layer pictures of the current picture into the prediction model, and generating the reconstructed data of the second picture component of the current picture,
wherein the one or more low time layer pictures are one or more pictures obtained by decoding the bitstream of the video before decoding the current picture according to a decoding sequence.

7. The method of claim 6 , wherein the prediction model comprises a denoising network and a prediction network, and the method further comprises:
inputting the reconstructed data of the first picture component of the current picture into the denoising network, and obtaining a first denoised data;
inputting the reconstructed data of the one or more low time layer pictures to the denoising network, and obtaining a second denoised data;
fusing the first denoised data and the second denoised data to obtain a fused picture feature; and
inputting the fused picture feature into the prediction network, and generating the reconstructed data of the second picture component.

8. The method of claim 7 , wherein fusing the first denoised data and the second denoised data to obtain the fused picture feature further comprises:
a first preprocessing, configured to resize the first denoised data or the second denoised data, such that a size of the first denoised data is the same as a size of the second denoised data after the first preprocessing.

9. The method of claim 1 , further comprising:
decoding the bitstream of the video to obtain model parameters of the prediction model; and
building the prediction model based on the model parameters.

10. The method of claim 1 , further comprising:
determining a training data set;
determining model parameters of the prediction model by training with the training data set; and
building the prediction model based on the model parameters.

11. The method of claim 10 , wherein determining the training data set comprises:
decoding the bitstream of the video, and determining training data identification information; and obtaining the training data set according to the training data identification information;
or,
obtaining the training data set according to preset training data identification information,
wherein the method further comprises: obtaining the training data set from a remote server indicated by the training data identification information.

12. An encoding method, applied to an encoder, the method comprising:
coding reconstructed data of a first picture component of a current picture;
determining identification information of the current picture in a video, wherein the identification information indicates whether to use a prediction model for determining reconstructed data of a second picture component of the current picture;
skipping encoding of the reconstructed data of the second picture component of the current picture, in response to the identification information indicating using the prediction model for determining the reconstructed data of the second picture component, wherein the second picture component differs from the first picture component; and
signaling encoded bits of the current picture into a bitstream of the video.

13. The method of claim 12 , wherein determining the identification information of the current picture comprises:
determining a first performance parameter and a second performance parameter of the current picture, wherein the first performance parameter represents encoding efficiency of a removed component, and the second performance parameter represents encoding efficiency of an un-removed component; and
determining the identification information according to the first performance parameter and the second performance parameter.

14. The method of claim 13 , wherein determining the identification information of the current picture comprises:
determining a time layer attribute of the current picture, wherein the time layer attribute indicates whether the current picture is a high time layer picture or a low time layer picture; and
determining the identification information according to the time layer attribute.

15. The method of claim 14 , further comprisin

Assignees

• Guangdong Oppo Mobile Telecommunications Corp Ltd

Inventors

• Yuan Hui
• Yang Ye
• Liu Yao
• Li Ming

Classifications

• H04N19/186 Primary

  the unit being a colour or a chrominance component · CPC title

• H04N19/105

  Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction · CPC title

• H04N19/136

  Incoming video signal characteristics or properties · CPC title

• H04N19/117

  Filters, e.g. for pre-processing or post-processing (sub-band filter banks H04N19/635) · CPC title

• H04N19/172 Primary

  the region being a picture, frame or field · CPC title