基于自适应背景匹配的运动目标检测算法  

Moving target detection algorithm based on adaptive background matching

作　　者：高振华[1] 黄山[1,2] 张洪斌[2] 于艳飞[3] 

机构地区：[1]四川大学电气信息学院,四川成都610065 [2]四川大学计算机学院,四川成都610065 [3]中国北车集团大连机车研究所有限公司技术中心,辽宁大连116021

出　　处：《计算机工程与设计》2013年第8期2839-2843,共5页Computer Engineering and Design

摘　　要：为了减小所提取的背景对后续目标检测的影响,提出了基于自适应背景匹配的运动目标检测算法。通过改进的Mode算法结合自适应背景匹配更新机制获得可靠的背景,在此基础上结合绝对差分和具有双动态阈值的Cauchy分布模型建立了目标检测总体模型,从而实现对视频序列中的运动目标进行检测。通过对行人和车辆等不同监控对象的实验,验证了算法具有良好的噪声抑制能力和对慢目标良好的鲁棒性,同时能有效的分割前景目标。To reduce the background extraction influence on handling moving targets,the moving target detection algorithm based on adaptive background matching is proposed.Firstly,the most confident background is obtained by using the improved Mode algorithm and adaptive background matching mechanism.After generating the adaptive background,the general model is constructed by using the absolute difference and Cauchy model with double dynamic threshold,and then the general model is used to detect the moving target in the video sequences.The experimental results of pedestrians and vehicle detecting show that this method is avoiding the noise,robustness for the slower moving target and effective for detecting.

关 键 词：Mode算法 背景提取 自适应更新 绝对差分 Cauchy模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]