Skip to Main Content
Texture synthesis is a technique of generate new large image, from a small sample, that different from the original but has the same visual appearance. This technique is very valuable for filling missing information in images. In this paper, we apply Markov Random Fields (MRF) based texture synthesis technology to fill in occlusion areas on remotely sensed images. In order to improve the quality of results, we use GIS vector data to help to fill in the occlusion regions. Experiments show that the proposed method can obtain satisfying results and is practical in remote sensing application.