基于统计特性与边缘特性融合的阴影检测方法

A Shadow Detection Algorithm Based on Statistical and Edge Properties

针对视频监视系统中的阴影检测问题,提出了一种基于灰度比值统计与边缘特征分析相融合的阴影检测方法.根据阴影的面积和灰度特征,采用灰度统计和正态分布曲线拟合的方法估计出阴影灰度的分布区域,实现对阴影的初分割.利用熵判据对初分割的结果进行判断,对于过分割的情况,依据目标与阴影间的位置关系,利用阴影的边缘特性进行目标重构.不同光照环境下的视频监控图像测试结果表明,该方法可以用于灰度图像和彩色图像,能够有效检测出阴影区域并保持目标的完整性.

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.