In the process of optical remote sensing image imaging, shadow areas often appear in remote sensing images due to the occlusion of sunlight by vegetation, buildings, clouds and other objects. These shadow areas often contain a lot of important information and targets, which have a great impact on ground target recognition, feature edge extraction, and surface information inversion. Especially as the resolution of remote sensing images has entered the sub-meter level in recent years, the need for efficient shadow compensation methods has become more and more urgent.In this paper,aiming at the problem of color distortion after the common shadow compensation algorithms,combining with the typical characteristic of half-shadow area and the concept of Homogeneous area,this paper improve the formula of Wallis fliter and research an improved shadow compensation algorithm based on Lab color space.Wipe off the boundary of shadow areas,keep the visual coherence of the image betterly.The experimental result shows that the algorithm can achieve a better performance without manual intervention.