Shadow detection and compensation in high resolution remote sensing images is very critical for target identification and information recovering. Based on a series of analysis about shadow properties, several properties were selected to compose together to detect shadow. Through improving Otsu method, more suitable thresholds of each property could be acquired automatically, which realized automatic detection of shadow. In order to highlight the information of objects inside shadow areas, an improved shadow compensation strategy based on the Wallis filter principle was proposed. Experiments indicate that the detection method is effective, and the accuracy and application range are improved, the compensation result is outstanding, and details sheltered by shadow also reappear.