Abstract:Concerning the problem that matching time and accuracy requirements can not be met the simultaneously in image matching technology, a method based on feature points matching was proposed. Landmark matching was achieved successfully by using Random Forest (RF), and matching problem was translated into simple classifying problem to reduce the complication of computation for real-time image matching. Landmark image was represented by Features from Accelerated Segment Test (FAST) feature points, the scale and affine invariability of FAST feature points were improved by Gaussian pyramid structure and affine augmented strategy, and the matching rate was raised. Comparing with Scale-Invariant Feature Transform (SIFT) algorithm and Speed Up Robust Feature (SURF) algorithm, the experimental results show that the matching rate of the proposed algrorithm reached about 90%, keeping the matching rate approximately with SIFT and SURF in cases of scale change, occlusion or rotation, and its running time was an order of magnitude than other two algorithms. This method matches landmarks efficiently and its running time meets the real-time requirements.
[1] ZITOVA B, FLUSSER J. Image registration methods:a survey[J]. Image and Vision Computing, 2003, 21(11):977-1000. [2] 饶俊飞.基于灰度的图像匹配方法的研究[D]. 武汉:武汉理工大学, 2005:14-19.(RAO J F. Research of image matching method based on intensity[D]. Wuhan:Wuhan University of Technology, 2005:14-19.) [3] 惠国保, 童一飞, 李东波. 基于改进的图像局部区域相似度学习架构的图像特征匹配技术研究[J].计算机学报, 2015, 38(6):1148-1161. (HUI G B, TONG Y F, LI D B. Image features matching based on improved patch similarity learning framework[J]. Chinese Journal of Computers, 2015, 38(6):1148-1161.) [4] MIKOLAJCZYK K, SCHMID C. A performance evaluation of local descriptors[C]//Proceedings of the 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2003:257-263. [5] LOWE D G. Distinctive image features from scale invariant key-points[J]. Computer Vision, 2004, 20(2):91-110. [6] 徐秋辉, 佘江峰, 宋晓群, 等. 利用Harris-Laplace和SIFT描述子进行低空遥感影像匹配[J].武汉大学学报(信息科学版), 2012, 37(12):1443-1447.(XU Q H, SHE J F, SONG X Q, et al. Matching low altitude RS image with Harris-Laplace and SIFT descriptor[J]. Geomatics and Information Science of Wuhan University, 2012, 37(12):1443-1447.) [7] BAY H, TUYTELAARS T. SURF:speeded up robust features[C]//ECCV 2006:Proceedings of the 9th European Conference on Computer Vision, LNCS 3951. Berlin:Springer, 2006:404-417. [8] 闫利, 陈林. 一种改进的SURF及其在遥感影像匹配中的应用[J]. 武汉大学学报(信息科学版), 2013, 38(7):770-773.(YAN L, CHEN L. A modified SURF descriptor and its application in remote sensing image matching[J]. Geomatics and Information Science of Wuhan University, 2013, 38(7):770-773.) [9] LEPETIT V, FUA J P. Keypoint recognition using randomized trees[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(9):1465-1479. [10] LEPETIT V, PILET J, FUA P. Point matching as a classification problem for fast and robust object pose estimation[C]//CVPR 2004:Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2004, 2:Ⅱ-244-Ⅱ-250. [11] LEISTNER C, SAFFARI A, BISCHOF H. MIForests:multiple-instance learning with randomized trees[C]//ECCV 2010:Proceedings of the 11th European Conference on Computer Vision, LNCS 6316. Berlin:Springer, 2010:29-42. [12] GALL J, YAO A, RAZAVI N, et al. Hough forests for object detection, tracking, and action recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(11):2188-2202. [13] ROSTEN E, DRUMMOND T. Machine learning for high-speed corner detection[C]//ECCV 2006:Proceedings of the 9th European Conference on Computer Vision, LNCS 3951. Berlin:Springer, 2006:430-443. BackgroundYANG Qili, born in 1993, M. S. candidate. Her research interests include image processing, pattern recognition.ZHU Lanyan, born in 1966, M. S., associate Professor. Her research interests include spatial data processing and error analysis, quality evaluation of GIS.LI Haitao, born in 1991, M. S. candidate. His research interests include spatial analysis of GIS, image processing.