Video Object Segmentation with Shadow Removal

Video segmentation is a key step in image sequence analysis and its results are extensively used for determining motion features of scene objects, as well as for compression. To overcome the problems of semiautomatic segmentation, it is necessary to develop an automatic segmentation algorithm that requires fast implementation without user assistance. These requirements are particularly important for real time applications. In this paper, moving segmentation algorithm using change detection based on background registration technique for video sequences is implemented. A background registration technique is used to construct constant background information from the video sequence. Gradient filter is used for shadow removal. Finally, a post-processing step is used to remove noise regions and produce a smooth shape boundary. Experimental results demonstrate that the proposed method can eliminate the noise region and thus achieves a significantly improved segmentation result. This algorithm makes use of the temporal information to achieve higher efficiency for video surveillance systems.