Research On Face Image Repair Technology

Image repair technology uses given information to not only restore missing details in a broken image with some certain rules,but also achieve visual effects.The paper focuses on the repair technology of face images.The main research work includes:1?According to the self-similarity of skin textures in the face image,this paper proposes an algorithm to complete the image repair which uses texture synthesis and priority strategy.Firsly,this algorithm calculates the priority of the area to be repaired based on the luminance lines and the confidence of the repair block.Secondly,performs a local search on the face by the calculated priority.Then finds the closest texture block,and completes the repair.The experimental results shows that,compared with the existing image processing software,this algorithm can achieve a better promance.2?In order to solve the problem of large-area restoration in the face images,this paper proposes a repair algorithm consisting of the generation of confrontation network and face semantic knowledge.The confrontation network contains a generator and two discriminators in this repair algorithm.For the semantic knowledge of the face image,the algorithm uses an encoder-decoder to compare two parsed results of the generated image and the original image,which can improve the repair quality.The experimental analysis shows that,this proposed algorithm can be suitable for the face image repair with large-area regional fouling.3?Based on the proposed research,the first face repair algorithm can be applied to the social face photo editing system by the form of a dynamic link library.This system adjusts the repair radius to fit for the different sizes of the face mask,and compilshes a better face rapair,which can greatly increase the face repair quality.