复杂背景下运动目标的提取  

Extraction of moving objects in clutter background

作　　者：张旭光[1,2] 韩广良[1] 孙巍[3] 雷凯[1,2] 王延杰[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所 [2]中国科学院研究生院,北京100039 [3]吉林大学测试科学实验中心

出　　处：《光电工程》2006年第4期10-13,54,共5页Opto-Electronic Engineering

摘　　要：提出了在复杂背景以及摄像机运动的情况下运动目标提取的算法。采用了新的局部投影的方法,在图像中合理地划分了投影区域,在求投影偏移量时提出了峰值匹配的算法,合理地检测到了投影偏移量。求得各个局部运动矢量后,采用出现频率最高的局部运动矢量代表全局运动矢量,成功的补偿了背景的运动。实验结果表明该算法在图像背景复杂且由于摄像机运动使背景发生明显变化的情况下,可稳定准确地提取出运动目标,并能有效克服背景中存在的运动干扰。A New algorithm is proposed to extract moving objects under complex background in the case of the movement of camera. A new local projection algorithm is used to plot the regions of projection in reason. In order to get the excursion of projection correctly, the peak value of the projection is applied to match in current frame. After the movement vectors of every region are gained, the biggest frequency of the local movement vectors is used to represent the whole vector, through which this we can equalize the movement of background successfully. Experimental results show that this algorithm can extract moving objects correctly under complex background though the camera is moving .This algorithm also can overcome the disturbance of other moving object.

关 键 词：目标跟踪 目标提取 复杂背景 灰度投影 

分 类 号：V556[航空宇航科学与技术—人机与环境工程] TP391[自动化与计算机技术—计算机应用技术]