基于ORB的快速完全仿射不变图像匹配

J4 ›› 2014, Vol. 36 ›› Issue (02): 303-310.

基于ORB的快速完全仿射不变图像匹配

侯毅,周石琳,雷琳,赵键

（国防科学技术大学电子科学与工程学院，湖南长沙 410073）

收稿日期:2012-07-12 修回日期:2012-10-31 出版日期:2014-02-25 发布日期:2014-02-25
基金资助:
国家自然科学基金资助项目（61105031）

Fast fully affine invariant image matching based on ORB

HOU Yi,ZHOU Shilin,LEI Lin,ZHAO Jian

College of Electronic Science and Engineering,National University of Defense Technology,Changsha 410073,China)

Received:2012-07-12 Revised:2012-10-31 Online:2014-02-25 Published:2014-02-25

摘要/Abstract

摘要：

ASIFT具有完全仿射不变性，但计算耗时；ORB实时性好，但仿射不变性差。为了在图像匹配中同时兼顾仿射不变性和实时性，利用模拟相机在不同视点下成像的手段使得ORB具备完全仿射不变性，进而提出了一种基于ORB的快速完全仿射不变图像匹配新算法(AORB)。首先通过模拟相机在不同视点下成像以获取模拟的图像，然后用快速的ORB算法对所有模拟的图像对进行匹配，最终取得完全仿射不变性。实验结果表明，该算法能够满足完全仿射不变图像匹配需求，并且相比基于OpenMP的ASIFT计算速度提高了约6倍。

关键词: ORB;ASIFT;AffineSIFT;仿射不变;图像匹配

Abstract:

Affine-SIFT is fully affine invariant but its computation is time-consuming. Oriented FAST and Rotated BRIEF (ORB) is extremely fast but is not affine invariant. In order to solve the problem that it is difficult to balance the good affine invariance and the real-time performance in image matching, a new fast method (AORB, Affine-ORB) for fully affine invariant image matching based on ORB, which applies the ASIFT method simulated in all views to obtain the full affine invariance of ORB, is proposed. Firstly, it simulates enough image views obtainable by varying the viewpoints of camera. Secondly, all simulated image pairs are matched using fast ORB. Thus, the full affine invariance is obtained. Experimental results show that proposed method is efficient in fully affine invariant image matching and it is 6 times faster than ASIFT.

Key words: ORB;ASIFT;affine-SIFT;affine invariant;image matching

侯毅,周石琳,雷琳,赵键. 基于ORB的快速完全仿射不变图像匹配[J]. J4, 2014, 36(02): 303-310.

HOU Yi,ZHOU Shilin,LEI Lin,ZHAO Jian. Fast fully affine invariant image matching based on ORB [J]. J4, 2014, 36(02): 303-310.

参考文献

［1］Mikolajczyk K, Schmid C. Scale and affine invariant interest point detectors［J］. International Journal of Computer Vision, 2004, 60(1):6386.
［2］Matas J, Chum O, Urban M, et al. Robust wide baseline stereo from maximally stable extremal regions［C］∥Proc of the British Machine Vision Conference, 2002:384393.
［3］Mikolajczyk K, Tuytelaars T, Schmid C, et al. A comparison of affine region detectors［J］. International Journal of Computer Vision, 2005, 65(1):4372.
［4］Morel J M, Yu G. ASIFT:A new framework for fully affine invariant image comparison［J］. SIAM Journal on Imaging Sciences, 2009,2(2):438469.
［5］Lowe D. Distinctive image features from scaleinvariant keypoints［J］. International Journal of Computer Vision, 2004, 60(2):91110.
［6］Rublee E, Rabaud V, Konolige K, et al. ORB:An efficient alternative to SIFT or SURF［C］∥Proc of 2011 IEEE International Conference on Computer Vision (ICCV 2011), 2011:25642571.
［7］Rosten E, Porter R, Drummond T. Faster and better:A machine learning approach to corner detection［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(1):105119.
［8］Calonder M, Lepetit V, Strecha C, et al. BRIEF:Binary robust independent elementary features［C］∥Proc of European Conference on Computer Vision, 2010:1.
［9］Calonder M, Lepetit V, Ozuysal M, et al. BRIEF:Computing a local binary descriptor very fast［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(7):12811298.
［10］Rosin P L. Measuring corner properties［J］. Computer Vision and Image Understanding, 1999, 73(2):291307.
［11］Moisan L, Stival B. A probabilistic criterion to detect rigid point matches between two images and estimate the fundamental matrix［J］. International Journal of Computer Vision, 2004, 57(3):201218.
［12］The standard Mikolajczyk database ［EB/OL］. ［20070715］. http://www.robots.ox.ac.uk/~vgg/research/affine/.
［13］Demo_ASIFT_src［CP/OL］. ［20110430］.http://www.cmap.polytechnique.fr/~yu/research/ASIFT/demo.html.