Mural Image Shedding Diseases Automatic Marking and Inpainting Algorithm

Murals are witnesses of Chinese history and culture, and have important historical research value. The research object selected in this article is the widely existing shedding disease. The traditional methods for repairing shed diseases usually include manual annotation and simultaneous repair of structure and texture, but this method has little effect on repairing shed diseases through contour lines. In addition, it is more difficult to repair diseases caused by the detachment of contour lines. To tackle these challenges, it is necessary to explore more suitable repair methods. Considering the particularity of shedding diseases caused by contour lines, this paper proposes an automatic marking and restoration algorithm for mural shedding diseases. We present a novel approach that incorporates enhancements across multiple stages, including multiscale morphological edge gradient detection, separate shedding edges and internal filling, relevant structural information restoration and textural information restoration. By calculating the image quality assessment (IQA) index to form a comprehensive evaluation system, our validation results demonstrate the effectiveness of the proposed method across multiple types of shedding disease. Notably, enhancements in edge detection and structural priority restoration optimization modeling significantly boost the whole method's performance. These improvements underscore the method's completeness, accuracy, and performance efficacy.