Abstract
This paper describes a method that can improve the visibility of features in shadowed regions in aerial images while retaining the quality of non-shadowed regions. The method first calculates the probabilities of each pixel belonging to shadowed class and non-shadowed class using color features from a smoothed image made with a morphological filter, and then compensates the intensity and saturation of each pixel according to the probability of its belonging to the shadowed class. Experimental results indicate that the accuracy of the proposed classification is about 95%, and that it improves the visibility of shadowed features while retaining the quality of non-shadowed regions.
