视频监视中运动物体检测的一种新算法

本文提出一种结合阈值分割以及区域生长的算法检测视频中运动物体的算法.首先对相邻的图像帧进行差分并根据3σ准则二值化差分图像,然后对二值图像进行扫描去除虚假目标区域和孤立亮点、并记录各区域的边界值,最后用区域生长的办法得到运动物体的完整信息.实验表明,在背景复杂、光照不均匀的视频中,该算法比帧间差分法、数学形态学的方法等能够更精确地检测出多个运动物体,在提取目标的同时,不留下任何背景像素,使下一步的目标跟踪更精确.

New algorithm of moving objects detection for video surveillance

A new algorithm for moving object detection based on threshold segmentation and region growing method is proposed in this paper. Firstly,we operate difference calculation to adjacent frames based on three-regulation,and do Binarization processing. Then the bilingual picture is scanned to remove the lonely polluted pixels and false objects,and the statistic of area boundary is record at the same time. Lastly,region growing method is used to get entire information of objects. Experiment indicates that the algorithm in this paper is more especially efficient in detecting multi moving objects in complicated and uneven lighting background compared to consecutive frames difference method and mathematic morphology method. No background pixels are left when extracting objects,so the measuring precise get improved in the next object tracing in this algorithm.