基于光流场技术的复杂背景下运动目标跟踪  

Moving target tracking based on optical flow field analysis in complex background

作　　者：李宏[1] 杨廷梧[1] 任朴舟[1] 李朝晖[1] 

机构地区：[1]中国飞行试验研究院,陕西西安710089

出　　处：《光电工程》2006年第10期13-17,共5页Opto-Electronic Engineering

基　　金：航空基金资助项目(03I30003)

摘　　要：针对光电经纬仪在飞机起飞着陆段跟踪角速度大、背景复杂的情况,根据小波多分辨率技术,提出利用小波分层组成低频图像金字塔,利用光流技术进行匹配解算,实现由粗到精的特征点匹配跟踪。通过对飞行试验视频图像进行事后自动跟踪试验,结果表明,该算法能够稳定可靠跟踪多特征点目标,具有鲁棒性。A novel automated method is developed to track aircraft takeoff and landing in a complex background. It involves three processes, Firstly, low frequency image pyramid is represented based on Wavelet multi-resolution and hierarchical theory. Then, an optical flow approach is used for matching of the pyramid representation. Finally the tracking method is completed through coarse-to-fine matching of feature point tracking. This method is evaluated by tracking flight video and found to track multiple targets with stability, reliability and robustness.

关 键 词：特征点跟踪 图像金字塔 小波多分辨率分析 光流场 目标跟踪 

分 类 号：V248[航空宇航科学与技术—飞行器设计]