面向飞行器平台的图像目标追踪算法  

Image Object Tracking Algorithm for Aircraft Oriented Platform

作　　者：柴天峰 朱小龙[2] 毛勇建[1] 卢永刚[1] 

机构地区：[1]中国工程物理研究院总体工程研究所,四川绵阳621900 [2]中国工程物理研究院,四川绵阳621900

出　　处：《现代防御技术》2017年第3期122-128,154,共8页Modern Defence Technology

摘　　要：以求解主动轮廓为目标观测手段,给出了基于卡尔曼滤波方法的可以追踪目标位置和目标在图像平面内面积的基础追踪算法。利用特征点匹配信息,解决了目标状态在图像平面内的急剧非线性变化导致基础追踪算法失败的问题。给出了结合特征点匹配与目标颜色统计直方图反向投影匹配2种方法的处理目标被遮挡情况的算法。通过实验验证,该算法在实验场景中,可以实现对目标位置和大小的追踪,可以成功地适应目标被遮挡的情况。Taking active contour as the target observation method, a basic tracking algorithm based on Kalman filtering method is presented for tracking the target position and computing the target area in the image plane. By using the matching information of the feature points, the problem that the sharp non- linear variation of the target state in the image plane leads to the failure of the basic tracking algorithm is solved. This paper presents an algorithm to deal with the occlusion of the target by two methods： the matching of feature points and the histogram of the target color. Experimental results show that the pro- posed algorithm can track the position and size of the target in the experimental scene, and can be suc- cessfully adapted to the situation that the target is occluded.

关 键 词：图像目标追踪 卡尔曼滤波 几何主动轮廓 特征点匹配 均值漂移算法 遮挡问题 

分 类 号：TN911.73[电子电信—通信与信息系统]