Image Stitching Algorithm Based on Region Division for Underwater Dam Crack Image

Abstract

The surface crack of underwater dam is one of the important indexes to evaluate the normal operation of the dam. Complete crack image is an important means to improve the accuracy of evaluation. In view of the limitations of traditional Algorithms in underwater crack image stitching, we propose an underwater dam surface crack image stitching algorithm based on region division(ISA-RD). First of all, an image enhancement algorithm aiming at increasing the number of feature points is used. Secondly, we simplify the process of feature point selection and matching by relying on the features of multiple regions in the local crack image, and improve the matching accuracy by mining the close relationship between the matching of feature points and different regions. Finally, the high matching feature point pairs are used for image fusion. We take the crack image of the real scene as the research object. Compared with the classical image stitching algorithm, the feature point matching algorithm proposed in this paper improves the accuracy of feature point matching. Obviously, the image quality after stitching is improved.