Automatic moving object extraction algorithm using histogram mapping and local watershed techniques for shadow effect

摘要

Background subtraction method is a typical technique used for real-time extraction of moving objects in video sequences. In this paper, we propose an efficient moving object extraction algorithm which is suitable for digital video monitoring applications. The proposed algorithm combines the histogram modification and the local watershed techniques. The present background image is estimated by sing the histogram modification table which maps the original background image to the present on. Then, the moving object region is extracted by background subtraction. Finally, the local watershed segmentation and hierarchical queue based region merging are applied on the uncertain region of moving object mask instead of whole image region, which can decrease processing time. Also proposed method is robust to both noise and intensity changes in the present input image. The proposed method is evaluated for real-time sequences and produces promising results.