Object-Based Shadow Extraction and Correction of High-Resolution Optical Satellite Images

Shadows in remote sensing images often cause problems, especially in land-cover classification and change detection. Hence, it is very useful if the radiance of shadow areas is corrected to the same level as that of shadow-free areas. In this study, a radiance measurement was carried out to investigate the spectral characteristics of sunlight. Then a method is proposed for shadow detection and correction of optical imagery. First, building shadow areas are detected using an object-based classification method that employs brightness values and their relationship with the neighboring area. Next, the detected shadow areas are corrected using a linear function to produce a shadow-free image. The shadow pixels with different darkness levels are corrected by using different ratios to obtain a smoothly restored image. The proposed semi-automated method was applied to a QuickBird and a WorldView-2 images of Tokyo, Japan, to demonstrate the effectiveness of the method.