红外序列图像目标跟踪的自适应Kalman滤波方法  

Moving object tracking based on adaptive Kalman filter

作　　者：高璐[1] 张大志[1] 田金文[1] 

机构地区：[1]华中科技大学多谱信息处理技术国防重点实验室,湖北武汉430074

出　　处：《红外与激光工程》2007年第5期729-732,757,共5页Infrared and Laser Engineering

基　　金：国家自然基金资助项目(60572048)

摘　　要：提出了一种用于动态序列图像目标跟踪的自适应Kalman滤波方法。该方法用函数估计的思想估计目标的当前运动模型,同时实时修改滤波器的统计模型,并将最小二乘支持向量机应用于对当前目标运动模型的估计。实验表明,此种改进的Kalman滤波器的算法在跟踪机动目标时具有良好的性能。Target tracking has been widely applied in military and civil fields. An adaptive Kalman filter algorithm is described for moving object tracking. Such approach uses function estimation strategy to estimate the movement models of the target. LS-SVM is applied to estimate the movement models of the target and the statistical models of the system. Experimental results show that the approach has a satisfactory tracking performance in case of moving object tracking.

关 键 词：目标跟踪 红外序列图像 KALMAN滤波 自适应滤波 运动模型估计 最小二乘支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]