一种改进的基于统计的阴影检测方法

阴影检测与抑制是视频处理系统不可或缺的环节。该环节将直接影响目标跟踪、物体识别等的精确性。常用的阴影检测算法很难同时解决好阴影的误检和漏检这一对矛盾问题。本文根据Gabor滤波器在空间域上可选带宽和方向的特性，并结合基于统计的方法来实现阴影检测。这种方法的优点是综合考虑了亮度、色差和纹理特性，从而能优化阴影检测效果。

An improved statistical shadow detection method

Shadow detection and suppression were key steps in video processing system. The accuracy of subsequent analysis, such as target tracking and object recognition, will be directly affected by this step. The paradox of the shadow detection was that it was difficult to solve false shadow detection and shadow missing problems simultaneously. This paper proposed an improved method to detect shadows. The method use Gabor filters to optimize traditional statistical method. The advantage of this method is considering brightness, color and texture features together. Experiments show that the good results obtained.