运动背景下的运动目标检测方法  

A Method for the Detection of Moving Target in Moving Background

作　　者：於时才[1] 吴键[1] 

机构地区：[1]兰州理工大学计算机与通信学院,甘肃兰州730050

出　　处：《计算机仿真》2011年第2期292-294,372,共4页Computer Simulation

摘　　要：对运动目标检测问题,提出了一种运动背景下的运动目标检测新方法,为了配准图像首先利用傅立叶变换的相位相关法进行背景补偿,然后采用六帧相邻图像差分相乘的方法增强运动目标的像素点,拉大目标与背景残留噪声的差异,利用区域-分裂合并法进行目标分割,将运动目标检测出来;从而计算出运动目标的形心坐标,并绘出运动目标的运动轨迹。由于方法稳定性强,准确率高的特点,可以使复杂背景能较好检测出运动目标,并且能够在正确分割的前提下可以很好绘出运动目标的运动轨迹。实验结果证明了算法的有效性和可行性。A method for the detection of moving target in moving background is proposed in this paper.First,the background is compensated with phase correlation of Fourier,and the pixels around the moving target are strengthened by multiplying the two difference images of six frames.The difference between the target and residual background noise are broadened.Then the target is segmented by region Split-merge algorithm,and finally the moving target is detected from the moving background.The centroid coordinates of moving target are calculated and the moving trace of moving target can be drawn out.Since this method is more stable and has higher accuracy,the moving target can be detected in the complex background and the moving trace of moving target can be drawn out with the support of right segmentation.Experiment results show that the algorithm is effective and feasible.

关 键 词：傅立叶变换 区域-分裂合并 运动目标检测 背景补偿 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]