基于CenSurE-star特征的无人机景象匹配算法

发布时间：2018-06-21 03:20

  本文选题：无人机景象 + CenSurE-star特征　； 参考：《仪器仪表学报》2017年02期

【摘要】：针对传统局部不变特征的景象匹配算法冗余点多、实时性差、抗几何变换不突出的情况,提出基于CenSurE-star的无人机(UAV)景象匹配算法。首先采用Cen Sur E特征星型滤波器(CenSurE-star)提取基准图和实时图中的特征点,并生成FREAK二进制描述符;然后将汉明距离作为特征点的相似性判定度量,采用K近邻距离比值的方法提取匹配点对;最后利用基于RANSAC的定位模型得到空间几何变换关系,实现图像匹配并获取定位点经纬坐标。算法性能评价实验表明,本文算法不仅相对于SIFT、SURF、ORB算法,对各种变换具有更好的鲁棒性,而且相对于改进的SIFT、SURF算法处理时间有更大程度的缩短,算法定位误差在0.8个像素内,尺度误差在0.02倍内,旋转角度误差在0.04°内。基于算法进行外场飞行实验,实验证明算法定位精度较高,可以适应地貌信息较少的环境,并能满足无人机视觉辅助导航的需求。
[Abstract]:In view of the traditional local invariant feature matching algorithm with more redundant points, poor real-time performance and anti geometric transformation, the image matching algorithm based on CenSurE-star (UAV) is proposed. Firstly, the Cen Sur E feature star filter (CenSurE-star) is used to extract the datum map and the feature points in the real-time map, and the FREAK binary description is generated. Then, the Hamming distance is used as the similarity determination measure of the feature point, and the matching point pair is extracted by the method of K nearest neighbor distance. Finally, the spatial geometric transformation relationship is obtained by using the RANSAC based location model, and the image matching is realized and the positioning point latitude and longitude coordinates are obtained. The algorithm energy evaluation experiment shows that the algorithm is not only relative to SIFT in this paper. The SURF, ORB algorithm has better robustness for various transformations, and compared with the improved SIFT, the processing time of the SURF algorithm is greatly shortened. The algorithm localization error is within 0.8 pixels, the scale error is within 0.02 times and the rotation angle error is within 0.04 degrees. The field flight experiment based on the algorithm shows that the algorithm has a higher positioning accuracy. It can adapt to the environment with less geomorphologic information and meet the needs of UAV vision aided navigation.
【作者单位】： 桂林电子科技大学电子工程与自动化学院;桂林航天工业学院无人机遥测重点实验室;
【基金】：国家自然科学基金(61361006) 广西自然科学基金(2015GXNSFBA139251) 广西自动检测技术与仪器重点实验室基金(YQ14203) 广西高校无人机遥测重点实验室主任基金(WRJ2015ZR02)项目资助
【分类号】：TP391.41;V279
，

本文编号：2046968