Information compensation on RS image based on a global optimization approach
Jianlin Gu, Renxi Chen, Yuande Jin · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2008
For information lost phenomenon on RS image, we present a robust completion algorithm based on a single image instead of adopting image placement measure. In order to avoid the occurrence of visually inconsistent results caused by greedy patch-by-patch manner, we creatively introduce GIS accessorial data to guide structure completion and pose this task in the form of a discrete global optimization problem. In the process of implementing this optimization scheme, our method integrates exemplar-based texture synthesis techniques and Dynamic programming algorithm. By this way, Information Compensation is separated into two independent processes to deal with. Comparative Experiments is conducted and prove the efficiency of our method successfully.