电子电气工程与控制

基于极线几何的统计优化特征匹配算法

  • 赵春晖 ,
  • 樊斌 ,
  • 田利民 ,
  • 胡劲文 ,
  • 潘泉
展开
  • 西北工业大学 自动化学院, 西安 710072

收稿日期: 2017-09-06

  修回日期: 2018-01-24

  网络出版日期: 2018-01-24

基金资助

国家自然科学基金(61473230,61603303);航空科学基金(2014ZC53030);陕西省自然科学基金(2017JM6027,2017JQ-6005);地理信息工程国家重点实验室开放基金(SKLGIE2015-M-3-4)

Statistical optimization feature matching algorithm based on epipolar geometry

  • ZHAO Chunhui ,
  • FAN Bin ,
  • TIAN Limin ,
  • HU Jinwen ,
  • PAN Quan
Expand
  • School of Automation, Northwestern Polytechnical University, Xi'an 710072, China

Received date: 2017-09-06

  Revised date: 2018-01-24

  Online published: 2018-01-24

Supported by

National Natural Science Foundation of China (61473230, 61603303); Aeronautical Science Foundation (2014ZC-53030); Natural Science Foundation of Shaanxi Province (2017JM6027, 2017JQ6005); State Key Laboratory of Geo-information Engineering under grant agreement (SKLGIE2015-M-3-4)

摘要

针对基于特征点的图像匹配中匹配数目不多以及重复结构下匹配较差等问题,提出了一种基于极线几何的统计优化特征匹配算法。利用正确匹配特征点之间满足对极约束的特点,从而可以减小特征点搜索区域,避免由于重复结构引起的误匹配对。首先使用一个小的特征点样本估计图像之间的基本矩阵,并利用它结合对极约束模型来引导特征匹配;然后利用基于特征点主方向和尺度信息的统计优化方法进一步消除误匹配,得到最终匹配结果。实验结果表明,该算法对图像的旋转和缩放变换具有良好的鲁棒性,匹配精度和数目有了很大提升,对于具有重复结构的图像匹配效果也较好。

本文引用格式

赵春晖 , 樊斌 , 田利民 , 胡劲文 , 潘泉 . 基于极线几何的统计优化特征匹配算法[J]. 航空学报, 2018 , 39(5) : 321727 -321727 . DOI: 10.7527/S1000-6893.2018.21727

Abstract

To solve the problems of small matching number and poor matching with repetitive structures in image matching based on feature points, a statistical optimization feature matching algorithm is proposed based on the epipolar geometry. As correct matching feature points satisfy the epipolar constraint, the feature points' search region can be reduced and the mismatches caused by repetitive structures can be avoided. Our approach first uses a small sample of features to estimate the fundamental matrix between images and leverages it for guiding feature matching based the epipolar constraint model. Then we use the statistical optimization method based on scale and main direction information of feature points to further eliminate mismatches and get the final matching results. Experimental results show that our algorithm also has good robustness against image rotation and scale transformation, and the matching precision and number have been greatly improved. Our algorithm is also effective for image matching with repetitive structures.

参考文献

[1] HARTLEY R, ZISSERMAN A. Multiple view geometry in computer vision[M]. Cambridge:Cambridge University Press, 2003:191-215.
[2] MUR-ARTAL R, MONTIEL J M M, TARDOS J D. ORB-SLAM:A versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5):1147-1163.
[3] ARYA S, MOUNT D M. Approximate nearest neighbor queries in fixed dimensions[C]//SODA'93:Proceedings of the Fourth Annual ACM-SIAM Symposium on Discrete Algorithms. Philadelphia, PA:Society for Industrial and Applied Mathematics, 1993:271-280.
[4] LOWE D G. Distinctive image features from scale-invariant key points[J]. International Journal of Computer Vision, 2004, 60(2):91-110.
[5] MUJA M, LOWE D G. Fast approximate nearest neighbors with automatic algorithm configuration[J]. International Conference on Computer Vision Theory and Applications (VISAPP), 2009, 2(1):331-340.
[6] KUSHNIR M, SHIMSHONI I. Epipolar geometry estimation for urban scenes with repetitive structures[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(12):2381-2395.
[7] ZHANG W, KOSECKA J. Generalized RANSAC framework for relaxed correspondence problems[C]//Third International Symposium on 3D Data Processing, Visualization, and Transmission. Piscataway, NJ:IEEE Press, 2006:854-860.
[8] SUR F, NOURY N, BERGER M O. Image point correspondences and repeated patterns[J]. Computer Vision & Pattern Recognition, 2011, 8(2):216-243.
[9] FISCHLER M A, BOLLES R C. Random sample consensus:A paradigm for model fitting with applications to image analysis and automated cartography[J]. Communications of the ACM, 1981, 24(6):381-395.
[10] CHIN T J, YU J, SUTER D. Accelerated hypothesis generation for multi-structure data via preference analysis[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(4):625-638.
[11] 王云丽, 张鑫, 高超, 等. 航拍视频拼图中基于特征匹配的全局运动估计方法[J]. 航空学报, 2008, 29(5):1218-1225. WANG Y L, ZHANG X, GAO C, et al. The global motion estimation method based on feature matching in video[J]. Acta Aeronautica et Astronautica Sinica, 2008, 29(5):1218-1225(in Chinese).
[12] 罗楠, 孙权森, 陈强, 等. 针对重复模式图像的成对特征点匹配[J]. 中国图象图形学报, 2015, 20(1):113-124. LUO N, SUN Q S, CHEN Q, et al. Pair-wise feature points based matching algorithm for repetitive patterns images[J]. Journal of Image and Graphics, 2015, 20(1):113-124(in Chinese).
[13] 刘威, 赵文杰, 李德军, 等. 一种基于ORB检测的特征点匹配算法[J]. 激光与红外, 2015, 45(11):1380-1384. LIU W, ZHAO W J, LI D J, et al. Feature points matching algorithm based on ORB detection[J]. Laser and Infrared, 2015, 45(11):1380-1384(in Chinese).
[14] SHAH R, SRIVASTAVA V, NARAYANAN P J. Geometry-aware feature matching for structure from motion applications[C]//2015 IEEE Winter Conference on Applications of Computer Vision (WACV). Piscataway, NJ:IEEE Press, 2015:278-285.
[15] SHAH R, DESHPANDE A, NARAYANAN P J. Multistage SFM:Revisiting incremental structure from motion[C]//20142nd International Conference on 3D Vision (3DV). Piscataway, NJ:IEEE Press, 2014:417-424.
[16] ZHANG Z, DERICHE R, FAUGERAS O, et al. A robust technique for matching two uncalibrated images through the recovery of the unknown epipolar geometry[J]. Artificial Intelligence, 1995, 78(1-2):87-119.
[17] 陈洁, 高志强, 密保秀, 等. 引入极线约束的SURF特征匹配算法[J]. 中国图象图形学报, 2016, 21(8):1048-1056. CHEN J, GAO Z Q, MI B X, et al. SURF feature matching based on epipolar constraint[J]. Journal of Image and Graphics, 2016, 21(8):1048-1056(in Chinese).
[18] 李立春, 张小虎, 傅丹, 等. 基于极线局部校正的特征匹配方法[J]. 光学技术, 2008, 34(2):285-288. LI L C, ZH X H, FU D, et al. Feature matching algorithm based on rectification of epipolar-line region[J]. Optical Technique, 2008, 34(2):285-288(in Chinese).
[19] 张培耘, 华希俊, 夏乐春, 等. 基于RANSAC算法的极线约束立体视觉匹配方法研究[J]. 组合机床与自动化加工技术, 2013(11):20-22. ZHANG P G, HUA X J, XIA L C, et al. Stereo matching with epipolar line constraints based on RANSAC algorithm[J]. Modular Machine Tool & Automatic Manufacturing Technique, 2013(11):20-22(in Chinese).
[20] 单欣, 王耀明, 董建萍. 基于RANSAC算法的基本矩阵估计的匹配方法[J]. 上海电机学院学报, 2006, 9(4):66-69. SHAN X, WANG Y M, DONG J P. The matching method based on RANSAC algorithm for estimation of the fundamental matrix[J]. Journal of Shanghai Dianji University, 2006, 9(4):66-69(in Chinese).
[21] 田谨思, 苏剑波. 基于Sampson误差计算单应的立体匹配[J]. 计算机工程与应用, 2005, 41(3):7-9. TIAN J S, SU J P. Stereo matching with epipolar line constraints based on RANSAC algorithm[J]. Computer Engineering and Applications, 2005, 41(3):7-9(in Chinese).
[22] STURM J, ENGELHARD N, ENDRES F, et al. A benchmark for the evaluation of RGB-D SLAM systems[C]//2012 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Piscataway, NJ:IEEE Press, 2012:573-580.
[23] AANAES H, DAHL A L, PEDERSEN K S. Interesting interest points[J]. International Journal of Computer Vision, 2012, 97(1):18-35.
文章导航

/