Object-Oriented Shadow Detection and Removal from Satellite Images

I introduce an object oriented shadow detection and removal method followed by an svm classification. In this technique, during image segmentation, shadow features are considered, suspected shadows are extracted with thresholding. Then, some dark objects which could be mistaken for shadows are ruled. Innerouter outline profile line (IOOPL) matching is used for shadow removal. The inner and outer outline lines of the boundary of shadows are taken as the IOOPLs. Shadow removal is performed using either Relative Radiometric Correction or polynomial fitting. Finally SVM classification is performed on input samples for improved accuracy.