基于改进的区域背景实时更新的目标检测  

Based on Improved Regional Background real-time Update Target Detection

作　　者：刘承祥[1] 葛广英[1] 潘至尊 张银苹[1] 

机构地区：[1]聊城大学物理科学与信息工程学院

出　　处：《聊城大学学报（自然科学版）》2013年第4期40-44,共5页Journal of Liaocheng University:Natural Science Edition

基　　金：国家自然科学基金资助项目(10874063)

摘　　要：针对静止摄像机条件下运动目标的检测问题,提出了基于改进的区域背景实时更新的目标检测算法,该方法首先对连续的三帧图像分别做帧间差分并将差分图像二值化,运用改进的线段编码的方法对二值化后的差分图像进行扩充以填补由帧间差分引起的空洞,然后用当前的二值化差分图像减去前一个二值化差分图像,差值为负的区域就是背景应该实时更新的区域,最后用传统的背景差分法就能检测出运动目标.实验结果表明,该方法不仅能在一直有运动物体的视野内获得完整的背景图像,而且背景的实时更新也能有效的减小噪声和突然进入摄像机视野的物体的干扰,有较强的鲁棒性.An target detection algorithm based on improved regional background real-time update has been proposed under the conditions of static camera motion detection. Initially, it differences the three consecutive frames and threshold the difference images respectively,fills the binary images＇ emp- ties caused by interframe difference through improved segment coding method; Secondly, after differen- cing the current binary images and the last one,it updates the background pixel erea lively which the negative result corresponds; Lastly,the motion target will be determined by the traditional background difference, the experimental results show that,this improved algorithm can obtain the background under the conditions that there are motion target ＇always, and also it can reduces the interference of noise and motion objects that enter the vision of camera suddenly,it has strong robustness.

关 键 词：区域背景 实时更新 目标检测 线段编码 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]