以扫描线为单位的背景提取与更新算法研究

Study on New Algorithm for Background Extraction and Updating in Unit of Scan Line

基于背景差分法的视频车辆监控系统实时性与准确性的设计要求,提出了一种新的背景提取与更新算法.该算法以扫描线为单位进行运算处理,简单易行.在背景提取和背景更新阶段,分别利用帧间差分和背景差分检测运动前景,再配合适当的运动区域判定条件,很好地消除了运动前景的影响,提高了准确性.通过将背景更新任务分解到多帧中处理,大大缩短了处理一帧图像的耗时,提高了系统实时性.实践证明该方法可行.

To meet the requirements for real-time and accuracy of a video-based vehicle monitoring system based on background subtraction, a new background extraction and updating algorithm is proposed. In the algorithm, both the extraction and updating in unit of scan line are carried out, which was simply and easy. In the stage of extraction and updating, moving objects by frame difference and background subtraction separately are detected, and influence with appropriate moving region criterions is removed, thus raised the accuracy. Through dividing updating tasks of one frame into several frames, greatly reduced the one frame processing time, thus raised the real-time property of the whole system. Experiments demonstrates that the algorithm is available for practice.