Image compression method based on adaptive downsampling and depth learning

Abstract

The invention discloses an image compression method based on adaptive downsampling and depth learning. The method mainly includes the following steps that: on an encoder side, a plurality of differentdownsampling modes and quantization modes are designed for a to-be-encoded original image, and then an optimal downsampling and quantization mode is selected from the plurality of modes by using a rate distortion optimization algorithm, and finally, the to-be-encoded image is subjected to downsampling and JPEG encoding in the selected optimal mode; on a decoder side, the decoded downsampling image is subjected to super-resolution reconstruction by adopting a convolutional neural network-based super-resolution reconstruction algorithm, and finally, a BM3D algorithm is used to further suppressthe residual compression effect after the super-resolution reconstruction, and a final decoded image is obtained. Experimental results show that compared with the mainstream encoding and decoding standards and advanced encoding and decoding methods, the proposed framework can effectively improve the rate distortion performance of encoded images and obtain better visual effects.

Description

technical field [0001] The invention relates to image compression and image coding technologies, in particular to an image compression method based on adaptive downsampling and deep learning, and belongs to the field of image communication. Background technique [0002] Most of the information in human activities is perceived through vision. As the carrier of visual information, image has the advantages of intuitive image, large amount of information, and easy understanding. However, in the actual image acquisition process, limited by transmission bandwidth and storage capacity, images are more or less lossy compressed. Among them, the mainstream compression coding standards applicable to still images include JPEG and JPEG2000, both of which were proposed by the Joint Photographic Experts Group in the early 1990s and early this century respectively. [0003] Due to its good compression performance and low complexity, JPEG has been widely used in the field of lossy image co...

AI Technical Summary

Problems solved by technology

However, most of the existing downsampling-based compression methods cannot effectively preserve image details in the process of downsampling and encoding, and there is still room for further improvement

Images