基于色度畸变和局部交叉熵的阴影消除

Shadow elimination by chrominance distortion and local kullback-leibler divergence

精确地提取运动目标是智能视频监控系统的核心任务之一。残留的阴影会导致目标形状的扭曲变形、目标相互连接等多种问题,这将给目标跟踪和识别等后续处理工作带来许多困难。根据阴影区域特有的光学属性,提出了一种基于色度畸变和局部交叉熵的阴影消除方法。

Extracting Moving objects accurately plays an important role in intelligent video-surveillance systems. Remnant shadows will bring many problems such as distortion of the object shapes and the unexpected adjacency between objects, and these problems can make the posterior work of object tracking and recognition very difficult. First, this paper analyse the optical characteristic of cast shadows, then propose a shadow elimination method based on chrominance distortion and local Kullback-Leibler divergence. At last, the experiment results are presented.