Detection and Enhancement of Shaded Areas in the High Resolution Satellite Data

Abstract

New high-resolution satellite data (IKONOS, QUickBird, SPOT5 and Eros) enable mapping in more detailed scales. The size of shadows is not of the order of pixel size any more. Misclassification of shadows in a high-resolution satellite image may lead to a shift of border between classes even by 30 meters. Shadows were grouped into five groups according to the shadowed surface and the type of the elevated object. Spectral analysis of these groups has been conducted and a simple method for detection of shaded areas has been proposed. This method utilizes map algebra. Shadows on vegetation and nonvegetation surfaces have been also distinguished. Three methods of image analysis to improve readability of shaded areas have been proposed: logarithmic enhancement, local contrast enhancement and ternary map. Objects hidden in shadow in the original image are better recognizable in the ternary map.