A Shadow Detection Algorithm Based on Statistical and Edge Properties
基于统计特性与边缘特性融合的阴影检测方法

Focusing on the shadow detection problem in video surveillance system,a new method based on gray scale ratio statistics and edge feature analysis was proposed.According to the area and gray characteristics of shadow,shadow′s gray distribution was estimated by gray statistics and normal distribution curve fitting method,and then the initial segmentation result was obtained.For the over segmentation result judged by the entropy criterion,the moving targets could be reconstructed based on the shadow′s edge characteristics gotten from the position between shadow and target.Under different illumination environment the experimental results indicate that the proposed algorithm could be applied in both gray image and color image and it could detect the shadow region effectively and maintain the integrity of the targets.