Research On Image Inpainting Based On Level Set Method

The damaged area must be inpainted in order to restore the damaged image and meet people's demand of high-quality printing. In this paper, a kind of technology on image inpainting based on level set is provided, which combined the technology of structure. The fruits of the research on the technology can be applied to such fields as archaeological fields, art, military, and advertisement. It can improve the quality of image inpainting and improve people's visual effect, and it especially has very important significance on protecting historical relic and restoring historical materials.This paper extends the image processing theory to image inpainting field. After analyzing the heuristic principles of the cognitive mechanism in the human visual systems, we apply the image restoring theory into image inpainting technology and establish the image inpainting model based on PDE and level set method according to principle frame of best guess theory and Bayesian structure system. We treat the missing regions as level sets and use the fast marching method (FMM) to propagate the image information. The boundary of the missing region is seen as 0-level set, and in the course of evolution, the region was inpainted. And we select appropriate weighting function to ensure the course is similar to the inpainting by hand.Although the inpainting basics are straightforward, most inpainting techniques published in the literature are complex to understand and implement. Our algorithm based on fast marching method is very simple to implement, fast, and produces nearly identical results to more complex, and usually slower, known methods. At last, we summary the problems of image inpainting at present and discuss the research direction in the future.