Coding and decoding method, image classification method and related device

The image is compressed through the spatial channel collaborative compression model, which solves the problem of poor image compression effect in the prior art, and realizes efficient compression and feature retention of large-scale images.

CN120201201APending Publication Date: 2025-06-24HUAWEI TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311787244.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-12-21
Publication Date
2025-06-24

AI Technical Summary

Technical Problem

The prior art has bottlenecks to improve the encoding and codec performance in image compression, especially for large-scale images that have poor compression effects and cannot meet the needs of multimedia applications.

Method used

Using a spatial channel co-compression model, the target tensor is entered into the model to obtain the spatial and channel compression tensors and compile them into the coded stream. The model is compressed by a convolutional layer. The spatial dimension of the spatial compression tensor is smaller than the target tensor and the channel dimension of the channel compression tensor is smaller than the target tensor, retaining the spatial and channel characteristics of the image.

Benefits of technology

It realizes efficient compression of images in spatial and channel dimensions, while retaining image features, improving the stability and efficiency of encoding and decoding.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120201201A_ABST
    Figure CN120201201A_ABST
Patent Text Reader

Abstract

The invention discloses a coding and decoding method, an image classification method and a related device, and belongs to the field of image compression. The method comprises the steps that a target tensor is input into a space channel collaborative compression model to obtain a compression tensor output by the space channel collaborative compression model, the compression tensor comprises a space compression tensor and a channel compression tensor, and the space compression tensor and the channel compression tensor are compiled into a code stream. According to the method and the device, the features of the target image in the channel dimension can be reserved while the target image is effectively compressed in the spatial dimension, and the features of the target image in the spatial dimension are reserved while the target image is effectively compressed in the channel dimension. On the basis, the space compression tensor and the channel compression tensor are compiled into the code stream, so that the code stream not only can effectively compress the target image, but also can fully reserve the characteristics of the target image in the space dimension and the channel dimension, thereby improving the stability of coding and decoding.
Need to check novelty before this filing date? Find Prior Art