移动背景下的运动目标跟踪  

Moving object tracking in the moving background

作　　者：梅跃松[1] 于剑桥[1] 陈曦[2] 

机构地区：[1]北京理工大学宇航学院,北京100081 [2]中国北方工业公司研究发展部,北京100053

出　　处：《红外与激光工程》2011年第4期757-761,共5页Infrared and Laser Engineering

基　　金：国防基础科研基金(B222006060)

摘　　要：运动目标跟踪在工业过程控制、医学研究、成像制导等领域具有重要的实用价值。目前的研究多基于背景静止的情况,对背景发生移动的情况研究较少。提出了一套完整的移动背景下的目标跟踪算法,首先使用基于互信息的方法配准序列图像的背景,然后使用差分的方法进行运动区域检测,并将其与图像分割技术相结合,得到目标跟踪模板。目标的跟踪基于Kalman滤波的预测,其匹配过程仍基于互信息理论,实验结果证明:该算法具有较高的计算效率和准确性,应用前景广泛。It is of important practical value to track moving objects in the fields of industrial process control,medical research,imaging guidance,etc.At present most research in this field are often based on static background,while those based on moving background are rather less.A complete algorithm of object tracking in the moving background was proposed.Firstly,the background of sequence images were registered by using the mutual information.Secondly,a differential method was used to detect moving object in combination with image segmentation technology to get the object template.Object tracking was based on Kalman filter and the object matching was still based on the mutual information theory.The experiments show that object tracking is rapid,accurate and robust by using this algorithm.Computational efficiency and accuracy of this new tracking algorithm is so high,and its application prospect will be broad.

关 键 词：目标跟踪 互信息 图像分割 卡尔曼滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]