Image Encoding and Decoding Methods for Boundary and Curvature Driven Inpainting

Abstract

The invention discloses an image coding and decoding method for boundary and curvature drive repairing. Firstly, an image subblock is divided into a structure block, a structure template block, a gradient block, a gradient template block and a non-characteristic block according to the edge intensity and the color variance. Due to the fact that the pixel value of the non-characteristic block cannot be depicted through a typical model, and meanwhile in order to effectively recover a structure subblock and a gradient subblock, the main structure and gradient parameters must be retained. Thus, the structure template block, the gradient template block and the non-characteristic block are coded with a traditional JPEG method, a linear interpolation method is used for repairing the gradient block, and finally, a diffusion model driven by the boundary interpolation and the curvature is used for recovering the structure block so as to obtain a decoded image. The defects that structure edges are fractured and the texture is vague in existing methods can be overcome.

Description

technical field [0001] The invention relates to a digital image compression method, in particular to an image encoding and decoding method for boundary and curvature driven restoration with high efficiency and more in line with the visual characteristics of human eyes. Background technique [0002] About 80% to 85% of the information acquired by the human perception system is visual information, so images and videos are the most important way for humans to receive information. In particular, the steady development of technologies such as the Internet, wireless communication, and mobile computing has enabled various types of multimedia services to continue to expand to various application fields in the social economy. People's consumption of image and video products and consumption expectations are higher than ever Much bigger. High-efficiency coding technology is an important guarantee for the successful promotion of many of the above applications. On the one hand, although...

AI Technical Summary

Problems solved by technology

However, the existing methods are still immature. Although they can be used to effectively decode the structural information of the image, the reconstruction results of the texture part have obvious artifacts, which affects the overall subjective quality of the decoded image.

Images