Perceptual video coding method based on area just-noticeable distortion

Abstract

The invention relates to a perceptual video coding method based on area just-noticeable distortion. The method comprises the following steps: obtaining all image blocks of each frame of image of a to-be-compressed video, obtaining a prediction JND threshold value of the image blocks through a trained JND prediction model, carrying out perception redundancy removal based on a target code rate and the prediction JND threshold value to obtain an optimal quantization parameter, and realizing perception video coding based on the optimal quantization parameter. Under the constraint that the subjective perception quality of the video is kept unchanged, under the condition of any target code rate, the code rate is saved to the maximum extent, and compared with the prior art, the method has the advantages of being low in complexity, high in robustness, high in efficiency and the like.

Description

Technical field [0001] The present invention relates to the field of video coding, in particular to a perceptual video coding method based on just perceivable distortions in regions. Background technique [0002] With the gradual increase in the ability of portable hardware devices to acquire rich multimedia, high-definition and 4K ultra-high-definition video have emerged. In order to facilitate the storage and transmission of large-capacity videos, it is necessary to further improve the video encoding performance. The High Efficiency Video Coding Standard (HEVC) proposed in 2012 has become the current mainstream advanced coding standard, but it still uses traditional objective evaluation standards to measure compression quality, such as mean square error (MSE) and peak signal-to-noise ratio (PSNR), etc. . However, such standards cannot accurately measure the subjective perception results of the human eye, because the human visual system (HVS) is different in sensitivity to dis...

AI Technical Summary

Problems solved by technology

However, most of the current JND models are built under the condition of a fixed code rate. When the target quantization parameters are updated, they need to be recalculated. It can be seen that the traditional JND models lack universality and high complexity; in addition, such models will JND The threshold is described as a continuous function of quantization parameters, and the latest research shows that the human eye has a stepwise perception of distortion, so the traditional JND model has certain limitations in simulating the perception process of HVS and guiding perceptual coding

Images