Abstract: A method to automatically detect and compensate shadows in high resolution remote sensing images is proposed. This method consists of initial shadow detection and post-procession. Through utilizing simple shadow spectrum signatures and complicated signatures shadows are detected automatically with relative automatic thresholds. Then the initial detected result is improved by a post-procession scheme based on the principles of mathematical morphology. Meanwhile an automatic compensating method which is essentially a mathematic model is explored, whose parameters are automatically acquired by calculating the mean and variance of illumination in shadow and non-shadow areas and using feature points extraction and matching strategy. Additionally, an integrated method using the compensation information from the whole shadow area and the small pixel window respectively is used to promote the final result. Experimental results indicate that both the multifarious conditions and post-processing can improve the accuracy of shadow detection results. The post-processing can also strengthen the completeness of shadow area. Moreover, the method of automatically acquiring the parameters in the compensation model is feasible and effectual. The integrated compensation strategy makes the brightness of shadow area heightened and the details and texture enhanced. In consequence, the real information shaded by shadow can emerge again.