Automatic Marking And Inpainting Of Cracks And Shedding Diseases On Ancient Building Mural Images

Ancient murals are a treasury of Chinese classical art,a witness to Chinese history and culture,and a precious historical and cultural heritage in China.Its historical research value is immeasurable.However,under the influence of external environment and human internal factors,most murals such as Mogao Grottoes,Western Thousand Buddha Caves and other mural groups have suffered from various kinds of diseases,including cracks,shedding,alkalizing,mildew and so on.In order to protect these treasures for a long time,it is an urgent and significant research content to inpaint accurately the damaged part of the mural.Compared with the traditional manual technology,the emergence of digital technology has brought new vitality to the inpainting of ancient building murals.The use of new technologies to mark and inpaint diseases can largely avoid the secondary damage to murals,and the excellent inpainting effect makes it an irreplaceable role in the analysis and development of historical and cultural heritage.Taking some mural images on Mogao Grottoes as an example,the two major diseases cracks and shedding,are marked and inpainted in this paper.Firstly,according to the linear structure characteristics of the mural cracks,the edge information of the cracks is obtained by mathematical morphology processing,and then the transformed image is segmented by adaptive threshold to ensure that the required pixels belong to the target area.The area is selected as the connected rule of the target area to remove the false target,the goal of accurate extraction is achieved,and the pixel of the damaged area is automatically marked.Aiming at the marked cracks,an improved Self-Organizing Map(SOM)image inpainting algorithm is adopted.The powerful function of SOM,which is unsupervised and automatic clustering,is used to complete the inpainting of mural cracks,and this algorithm provides an effective method for the inpainting of ancient building diseases.Secondly,different from the general shedding disease,the research object selected in this paper is the shedding disease passing through the outline.In the process of marking the shedding disease,the color characteristics of the shedding region are analyzed,and the edge information is extracted by multi-scale morphological edge gradient detection,then the target region of the image is highlighted by image enhancement technology,and the shedding edge is obtained.The mask of the shedding edge is filled internally to realize the marking of the shedding disease in the end;considering the particularity of the shedding disease,this paper proposes an image inpainting algorithm based on Genetic Algorithm(GA).The algorithm first emphasizes the restoration of structural information,then fills in the texture to obtain the ideal inpainting effect.To sum up,this paper focuses on the problems of the marking and inpainting of the cracks and shedding diseases in the murals of ancient buildings.From the perspective of digital image features,we uses mathematical morphology method to mark the mural diseases.The SOM neural network algorithm and Genetic Algorithm are used to inpaint the marked diseases,the performance of the inpainting is analyzed for the algorithm,the experimental results proves the feasibility and effectiveness of the algorithm in this paper finally.