Research On Thangka Image Inpainting Of Comprehensive Partial Differential Equations

Image inpainting is not only an important research topic in the fields of image restoration but also a research hotspot in image processing and computer vision.It is mainly used for heritage conservation, restoration of old photographs, removal of occlusions and so on,but heritage conservation is a very valuable application.Taking Thangka image as research object,the paper makes some study on the automatic selection of digtal image inpainting algorithms by studying the inpainting algorithms based on PDE and the exemplar-based inpainting algorithms in depth and combining the shape of damaged block,it's adjacent information with the characteristics of these algorithms.The paper mainly contains the following four parts.Firstly,we validated the classical algorithms in the fields of image inpainting and made an improvement for the defects of the exemplar-based inpainting algorithm and TV model.Secondly,the classification of the adjacent blocks of damaged block was studied,and damaged image is segmented by watershed algorithm in order to facilitate access to the adjacent information of damaged block. Thirdly,the classification was studied,which mainly contains the following three steps:(1) the second-order statistical information of the adjacent blocks was computed by GLCM,(2)Textures features of the adjacent blocks were extracted by GLCM and were normalized by Gaussian Normalization,(3)the adjacent blocks were classified into texture and non-texture by KNNC.Finally,we set the rules that are used to choose inpainting algorithms by combining the shape of damaged block,it's adjacent information with the characteristics of these algorithms,and implemented the automatic selection of inpainting algorithms.