动态背景下的背景运动补偿算法研究  

Research on Background Motion Compensation Algorithm in Dynamic Background

作　　者：赖飞 LAI Fei(College of Computer Science, Sichuan University, Chengdu 61006)

机构地区：[1]四川大学计算机学院

出　　处：《现代计算机（中旬刊）》2018年第3期74-77,共4页Modern Computer

摘　　要：背景运动的存在使得在动态背景下进行运动目标检测的难度大大增加,常见的帧间差分、背景减除等方法不能得到准确的检测结果,因此在使用传统方法前消除背景运动带来的影响显得尤为重要。针对背景运动参数估计过程中涉及的特征点匹配、外点去除以及参数的鲁棒性估计做出重要研究,提出一种精确度较高的背景运动参数估计方法。通过实验对比可以看到背景补偿后检测到的运动目标准确度有明显改进。The existence of background motion makes it more and more difficult to detect moving objects in the dynamic background. Common meth ods such as inter-frame difference and background subtraction cannot obtain accurate detection results, so the effect of removing background motion before using the traditional method appears particularly important. Studies the matching of feature points, the removal of external points and the robustness estimation of parameters involved in the estimation of background motion parameters, and proposes a more accurate background motion parameter estimation method. Through experimental comparison, it can be seen that the accuracy of the moving object detected after background compensation has been significantly improved.

关 键 词：运动背景 运动目标检测 特征点匹配 外点去除 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]