基于帧差法和背景差法的运动目标检测  

Moving Object Detection Based on Method of Frame Difference and Background Subtraction

作　　者：张应辉[1] 刘养硕 

机构地区：[1]东北大学计算机科学与工程学院,辽宁沈阳110000

出　　处：《计算机技术与发展》2017年第2期25-28,共4页Computer Technology and Development

基　　金：教育部在线教育中心项目(2016YB158)

摘　　要：随着视频监控的发展,智能监控算法也得到逐步的改进。智能监控中主要用到的是运动目标检测算法。在运动目标检测算法中,传统算法存在检测结果不准确、抗干扰性能低等缺点。为提高运动目标检测的准确性和高效性,基于三帧差法和背景差法,提出了一种二者相结合的运动目标检测算法。该算法利用三帧差法和背景差法分别获得前景目标图像,再通过或运算将这两幅前景图像合并成最终的前景图像。实验采用交通录像为实验样本,对视频中移动的车辆进行移动侦测。实验结果表明,该算法相较于其他算法更能检测出完整的车辆轮廓。从数值上分析也可以看出,该算法能够有效减少目标图像中的噪声,精确而快速地检测出运动目标,具有良好的鲁棒性。With the development of video surveillance,intelligent monitoring algorithm has been gradually improved. The moving target algorithm is mainly used in intelligent monitoring.In the moving target detection algorithm,the traditional algorithm is not accurate and lowin the anti jamming performance.In order to improve the accuracy and efficiency of moving target detection,a moving target detection algorithm is put forward based on combination of three frame difference method and background subtraction method. It uses three frame difference method and background subtraction method to obtain the foreground target image which are merged into the final foreground image then.The experiment is carried out using the traffic video as the experimental samples,and the moving vehicles in the video are detected.According to the experimental results,the algorithm can detect the more complete vehicle contour compared with others. From the numerical analysis,it can be seen that the algorithm can effectively reduce the noise in the target image,and detect the moving target accurately and quickly,with good robustness.

关 键 词：视频监控 三帧差法 背景差法 运动目标检测 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]