基于分级自适应背景差分的运动目标检测

针对经典自适应背景差分法中存在的问题,提出一种基于分级自适应背景差分的运动检测方法.方法结合了帧间差分法和背景差分法的优点,解决了运动检测中常见的拖尾问题和空洞问题.利用识别信息对感兴趣目标单独使用更新系数进行更新,在快速跟踪背景变化的前提下,能够保持对感兴趣目标的持续关注.

Moving Object Detection Based on Differentiated Self-adaptive Background Difference

In view of the faults in the classic self-adaptive background difference algorithm, an algorithm based on the differentiated self-adaptive background difference was presented. Combining the advantage of frame difference and background difference algorithm, the tailed and inane problems were resolved. Take advantage of the information of recognition, particular coefficient was used for the object of interest. In the premise of tracking background changes quickly, the algorithm can maintain the sustained attention of the interested object.