Automatic crack detection in heritage site images for image inpainting

Researchers have developed an automated method for detecting cracks in heritage structures, addressing a key challenge in digital restoration. The approach leverages order-statistics filtering and tensor voting to enhance crack contrast and curveness, respectively, followed by image dilation to generate target regions for inpainting algorithms. This innovation eliminates the need for manual region selection, streamlining the restoration process. Experiments conducted on images from Hampi, a UNESCO World Heritage Site in Karnataka, demonstrated the method's effectiveness in identifying and restoring degraded areas caused by aging or vandalism.

Turakhia, N., Shah, R., & Joshi, M. (2012). Automatic crack detection in heritage site images for image inpainting. Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing, 1–8. https://doi.org/10.1145/2425333.2425401