电光与控制, 2019, 26 (1): 34, 网络出版: 2019-01-19  

基于两层策略的SAR图像局部不变特征点的匹配方法

Matching of Local Invariant Feature Points in SAR Image Based on Two-Tier Strategy
作者单位
火箭军士官学校, 山东 青州 262500
摘要
在基于局部不变特征的SAR景像匹配制导中, 局部不变特征点的匹配实时性具有十分重要的工程意义。提出了一种基于两层策略的特征点匹配方法。首先根据特征点响应的阈值对实时图和参考图分别提取一级特征点和二级特征点, 然后依据二级特征点到一级特征点的距离使二级特征点隶属于距离最近的一级特征点, 由此达到对二级特征点进行分组的目的。进一步, 通过对一级特征点进行方向梯度描述, 借助kd-tree最近邻方法(NN)匹配策略实现特征点初步配对, 并结合RANSAC算法验证剔除误匹配点。最后, 对隶属于提纯后一级特征点的二级特征点进行BRIEF描述, 分组对二级特征点进行匹配。仿真实验结果表明, 特征点匹配速度大幅提升。
Abstract
In SAR scene matching guidance system based on local invariant feature, the real-time performance of the matching of the local invariant feature points is of great engineering significance.A new scene matching algorithm based on two-tier matching strategy is proposed to improve the speed of scene matching algorithm.First, the first-level and the second-level feature points are extracted respectively from the real-time and reference images according to the threshold value.Then, make each of the second-level feature point subordinate to the closest first-level feature point according to their distance.Further, the first-level feature points are described by SIFT descriptor and using nearest neighbor method based on kd-tree to achieve the initial matching of the first-level feature points.RANSAC algorithm is utilized to get rid of the error matches.Finally, the second-level feature points which are subordinate to the first-level feature points after eliminating the false matching points are described by BRIEF descriptor and then are matched.Experimental results show that the matching speed of feature points is improved.

赵爱罡, 范小虎, 赵乾, 王建永, 葛春. 基于两层策略的SAR图像局部不变特征点的匹配方法[J]. 电光与控制, 2019, 26(1): 34. ZHAO Ai-gang, FAN Xiao-hu, ZHAO Qian, WANG Jian-yong, GE Chun. Matching of Local Invariant Feature Points in SAR Image Based on Two-Tier Strategy[J]. Electronics Optics & Control, 2019, 26(1): 34.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!