An Effective Approach for Shadow Detection and Removal of Remote Sensing Images Using an Unified Framework

In remote sensing images, shadows created by big tower, bridges, tree. It may destroy the information of an image. Missing information has a direct effect on common analysis and processing of image, this leads to inefficient classification. To separate shadow and non shadow regions, morphological feature is consider for segmentation. Relevance Vector Machine (RVM) classifier can also classify both the shadow and non shadow regions. Then, the borders extraction is performed for extracting shadow part of that image. In the removal process all the shadow areas remain some unusual manner. To remove shadows, mean and standard deviation for the shadow and non shadow region of an image are calculated. Then, the mean difference between shadow and non shadow part of the image is applied to shadow part of an image by normalization.