RECONSTRUCTION OF MISSING DATA IN VHR IMAGES USING BANDELET AND EXEMPLAR BASED INPAINTING STRATEGIES

The process of inpainting is to reconstruct the lost or deteriorated parts of images. The reconstruction of missing data in very high spatial remote sensed image is of great challenge due to its complexity. Remotely sensed Very High Resolution (VHR) images may be partly contaminated by cloud and accurate reconstruction of such images is of great problem. In this paper two strategies were proposed one for cloud removal and other for reconstruction of missing regions to improve the accuracy. The cloud removal problem is achieved by means of bandelet transform and multiscale geometrical grouping algorithm. This algorithm aims at removing the cloud contaminated portions from the VHR images. Then Exemplar based inpainting strategy (approach) is used for reconstruction of missing regions. The two algorithms used here are very simple and efficient results can be obtained.