Image Inpainting for Paint Crack by Space Distance and Difference of Bilateral Filter

Abstract

This paper proposes a new inpainting method which is aimed at virtual restoration for paint crack. Target images of this research include missing regions which are models for high-density and widespread crack and paint flaking which is beyond a certain size. In conventional research, there is an effective method to restore these missing region. However, it's accuracy of texture extraction and texture selection are not enough. Proposed method improved the problem of conventional method that is including edge in texture by using difference of bilateral filter for texture extraction. Additionally, the accuracy of texture selection improved with use of inpainting method based on space distance. Effectiveness of proposed method can be confirmable by evaluating restored images by RMSE in both color and texture.