Shadow Removal from Remote Sensing Images by Efficient Inpainting Method

Abstract

Remote sensing images from satellite are useful for many different applications like change detection, identification of topological components shadow. Shadows have to be removed from the satellite images before it has to be processed for feature extraction. This paper proposes an efficient approach for shadow detection and removal in remote sensing images on urban areas. In our proposed work the image is first converted into HSV color model, shadow is detected by threshold based on Otsu’s method and subsequent boundaries are extracted. For shadow removal, inner-outer profile line (IOPL) is generated with respect to obtained boundaries obtained. Then shadow regions are masked, in painting algorithm is implemented to restore the original image. Thus new method accurately detects and removes shadow in remote sensing images.