基于对称差分和背景减的运动目标检测  

Moving object detection algorithm based on symmetrical-differencing and background subtraction

作　　者：邱光能 全惠敏[1] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082

出　　处：《计算机工程与应用》2014年第13期158-162,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.60835004)

摘　　要：通过对视频运动对象特点的分析,提出一种针对静态场景的运动目标检测算法。该算法采用一种改进的时间平均法初始化背景,在有目标的情况下也能构建出可靠的背景,并融合背景减法和多重对称差分法对背景进行自适应更新。实验结果证明,该算法计算简单,对光线变化具有适应性,能够完整地提取运动目标,改善了运动目标的检测效果,具有一定的鲁棒性。An algorithm for moving object detection with a static background is proposed by analyzing characteristic of movement objects in video. The algorithm uses improved time-averaged method to initialize background, it can obtain reliable background in the condition of having moving objects. During background updating, the paper combines background subtraction method and multi-symmetrical-differencing. Experimental results show that the proposed algorithm is quick, robust, adaptive to illumination changes and it improves the effect of moving object detection.

关 键 词：运动目标检测 自适应背景 多重对称差分 光线突变 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]