测绘通报 ›› 2018, Vol. 0 ›› Issue (7): 38-42.doi: 10.13474/j.cnki.11-2246.2018.0206

• 学术研究 • 上一篇    下一篇

结合SURF算法和单应性矩阵的无人机影像匹配

王晓红1, 邓仕雄2, 何志伟2, 曹留霞2, 闫星光2   

  1. 1. 贵州大学林学院, 贵州 贵阳 550025;
    2. 贵州大学矿业学院, 贵州 贵阳 550025
  • 收稿日期:2017-09-15 出版日期:2018-07-25 发布日期:2018-08-02
  • 作者简介:王晓红(1970-),男,博士,副教授,主要研究方向为3S技术应用和遥感图像处理。E-mail:gzdxwxh@163.com
  • 基金资助:
    贵州省自然科学基金(黔科合J字〔2014〕2070);贵州省科技计划课题(黔科合LH字〔2014〕7649);贵州省研究生教育教学改革重点课题(黔教研合JG字〔2015〕010);贵州大学测绘科学与技术研究生创新实践基地建设项目(贵大研CXJD〔2014〕002);贵州大学研究生重点课程建设项目(贵大研ZDKC〔2015〕008)

Study of UAV Image Matching Based on SURF Algorithm and Homography Matrix

WANG Xiaohong1, DENG Shixiong2, HE Zhiwei2, CAO Liuxia2, YAN Xingguang2   

  1. 1. Lin College, Guizhou University, Guiyang 550025, China;
    2. Mining College, Guizhou University, Guiyang 550025, China
  • Received:2017-09-15 Online:2018-07-25 Published:2018-08-02

摘要: 针对无人机影像受拍摄条件影响或区域环境复杂造成的匹配效果不佳,局部区域甚至无法匹配的问题,基于SURF算法,利用多重约束条件改进算法对无人机影像进行了特征匹配。该匹配算法首先利用SURF算法检测影像特征点,利用FLANN快速搜索结合KNN算法筛选特征点,选出构造单应性矩阵的最优内点匹配对,然后利用基于单应性矩阵的RANSAC算法过滤掉错误匹配。试验结果表明:与基于SURF算法的单一约束条件的无人机影像匹配相比,多重约束条件的无人机影像匹配算法在匹配质量优化的同时能提高无人机影像匹配集数量,该算法在误匹配减少的前提下能获得更多准确的特征点。

关键词: SURF算法, 无人机影像, 单应性矩阵, 极线约束, 特征匹配

Abstract: For unmanned aerial vehicle (UAV) video shooting conditions,or by the regional complex environment caused by matching the effect not beautiful,local area can't even match,this article proposes an improved algorithm using multiple constraints of UAV image feature matching based on SURF algorithm.The matching algorithm firstly uses SURF algorithm to detect image feature points,uses FLANN fast search algorithm combined with the feature of KNN algorithm screening point to select the structure one interior point matrix of the optimal matching of,then uses RANSAC algorithm based on single matrix to filter mismatch.The experimental results show that with a single constraint conditions based on SURF algorithm compared to the UAV image matching,UAV image matching algorithm of multiple constraints in optimization matching quality at the same time can improve the UAV image matching set number,under the premise of the algorithm in matching error reduce can obtain more accurate feature points.

Key words: SURF algorithm, UAV image, homography matrix, epipolar constraint, feature matching

中图分类号: