Image compression method and device, electronic equipment and storage medium

Abstract

The invention provides an image compression method and device, electronic equipment and a storage medium, and the method comprises the steps: giving an image generation network and a two-dimensional multi-channel random array with a low number of rows and columns, carrying out the iterative training of an initial image generation network based on a to-be-compressed image and the random array, enabling the image generated by the generative network to approach the target image, and finally determining network parameters of the image generative network; and representing the network parameters of the image generation network and the random number group as compression information of the to-be-compressed image. According to the method, the initial image generation network is iteratively trained based on the to-be-compressed image and the random number group, so that the image generation network obtained through training can output the generation image approaching the to-be-compressed image, and the network parameters of the image generation network and the random number group are used as the compression information representation of the to-be-compressed image; therefore, the to-be-compressed image can be compressed at a large compression degree, and the restoration precision of the image can be ensured.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image compression method, device, electronic equipment and storage medium. Background technique [0002] Image compression is a classic image processing application. Traditional image compression methods usually compress and save the image data itself or the coefficients of the data in the transform domain during encoding, and use inverse transform to decompress and restore the image according to the coefficients during decoding. [0003] For example, the traditional lossy image compression method is to remove part of the color value information of the image (such as JPEG or JPEG2000 format image), transform it into the frequency domain or wavelet domain, and then remove part of the transform domain coefficients and improve the encoding method, so as to achieve the overall compression . This method relies on prior knowledge and experiments of a large number ...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides an image compression method, device, electronic equipment and storage medium to solve the defects in the prior art that image compression relies on prior knowledge and the loss of image restoration is large when the image compression degree is too high

Images