一种基于改进三帧差分和ViBe算法的运动目标检测算法  

Moving target detection algorithm based on the improved three- frame differences and ViBe algorithm

作　　者：谢红[1] 原博 解武[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《应用科技》2016年第6期46-52,共7页Applied Science and Technology

基　　金：黑龙江省自然科学基金项目(F201339)

摘　　要：近几年提出的ViBe算法具有运算速度快,目标提取准确率高的特点,但是它对光照的突然变化也非常敏感,而且由于它的模型初始化的方式,也容易形成"拖影"现象。文中先通过引入传统三帧差分算法难以提取出完整的目标轮廓这一问题,提出一种结合边缘检测的改进三帧差分算法。该算法在提取出完整的运动目标的同时不会附加冗余的边缘信息。然后将该算法引入到ViBe算法中,使2种方法结合,一方面消除"拖影"现象,另一方面可以自适应地选择检测的方法,以消除光照突变造成的影响。试验结果表明,该改进算法是一种适应性强、鲁棒性高的运动目标检测算法。The ViBe algorithm proposed in recent years has the characteristics of high computing speed and high accuracy of object extraction. But it is sensitive to the light mutation. And it is also easy to form the ＂ smear＂ phenomenon because of its model initialization modes. Firstly,this article proposed the improved three-frame difference algorithm combined with edge detection by introducing the problem that the traditional three-frame difference algorithm is difficult to extract the complete contours of the targets. This algorithm can extract the complete moving targets without additional redundancy edge information. Then this algorithm is introduced into the ViBe algorithm and is combined with it. The improved algorithm can eliminate the ＂ smear＂ phenomenon. and it can select the suitable detection method adaptively to eliminate the impact caused by the light mutation. The experimental results show that,this method is an effective moving target detection method.

关 键 词：视觉背景提取算法 三帧差分法 边缘检测 “拖影”现象 光照突变 

分 类 号：TN911.73[电子电信—通信与信息系统]