Shadow Removal from Images Using an Improved Single-Scale Retinex Color Restoration Algorithm

Abstract

In this paper, we propose an improved single-scale retinex (SSR) color restoration algorithm to remove shadow from single color image. We use SSR to process the shadow and nonshadow regions first. Then introduce an approach based on deriving illumination invariant grayscale image to identify the shadow edge regions and process them by designing a nonisotropic low-pass filter. We combine the outputs and restore color information at the end. Our method can effectively remove shadow from a single color image and avoid intrinsic disadvantages. As a preprocess step of some computer vision tasks, such as image segmentation, face recognition and motion tracking, our algorithm plays an important role.