摘要
针对红外图像信噪比低,对比度不高,目标跟踪困难等复杂情况,提出了一种基于尺度空间特征点匹配的红外目标跟踪算法。设计出一种基于尺度因子变化的高斯卷积模板尺寸自适应调整以及连续帧目标区域双向配准的办法,前帧目标区域特征点匹配出后续帧目标区域特征点,以后者特征点集质心坐标去建立目标跟踪搜索窗,完成在视觉效果差的背景下的红外目标跟踪。实验表明,基于尺度空间特征点匹配的红外目标跟踪算法鲁棒性强,实时性高,能快速而有效地完成对红外目标跟踪。
In order to solve the problems of IR image＇s low SNR and low contrast, and difficulty of IR target tracking, a new IR target tracking algorithm based on scale space feature points matching was proposed. The size of template could automatically adjust along with the scale factor changing, and a bidirectional feature points matching algorithm of targets in sequential frames was proposed. After the feature points of a frame matching that of next frame in the target area of IR image, the center of latter feature points was taken as the center from which the target searching windows were set up continually to track the IR targets under the bad visual condition. By the IR target tracking algorithm based on scale space feature points matching, the robustness and real-time performance were improved. Experiments show that the IR targets can be tracked quickly and accurately by the proposed algorithm based on scale space feature points matching.
出处
《红外与激光工程》
EI
CSCD
北大核心
2011年第11期2104-2109,共6页
Infrared and Laser Engineering
基金
教育部中央高校基本科研业务费专项基金(JUSRP211A36
JUSRP111A41)
国家自然科学基金多年期重大国际合作研究项目(60910005)
关键词
尺度空间
红外目标
搜索窗
特征点匹配
scale space
IR target
searching window
feature points matching
作者简介
孔军（1974-），男，博士生，主要从事机器视觉和目标跟踪等方面的研究工作。Email：j—kong@163．com
导师简介：汤心溢（1970-），男，研究员，博士生导师，博士，主要从事成像系统技术、视频信号处理、红外动态仿真等方面的研究工作。Email：qq27@mail．sitp．ac．cn
