Improved matching point purification algorithm mRANSAC

被引：0
作者：
Wang, Yawei [1 ]
Xu, Tingfa [1 ]
Wang, Jihui [1 ]
机构：
[1] School of Optoelectronics, Beijing Institute of Technology
来源：
Dongnan Daxue Xuebao (Ziran Kexue Ban)/Journal of Southeast University (Natural Science Edition) | 2013年 / 43卷 / SUPPL.I期
关键词：
Image matching; Matching point purification; RANSAC (random sample consensus); Transformation matrix;
D O I：
10.3969/j.issn.1001-0505.2013.S1.034
中图分类号：
学科分类号：
摘要：
To sove the problem that correct match points cannot be effectively extracted in the match point purification links of current image matching algorithms, mRANSAC (multi-RANSAC) multi-transformation matrix method is proposed. Matching points cannot accurately correspond to each other due to the digital image's discrete sample style. There exist intrinsic position errors, and the corresponding transform matrices are different. Therefore, one transformation matrix cannot contain all the correctly matched points. The research results of RANSAC show that in the sets of non-maximal inner point number, enough points can induce correct matches, which can also be confirmed by the fact that different objective images result in different match numbers. Therefore, multi-transformation matrix is used to increase the matching point number and improve the purification efficiency. Three strategies, the set union method, the set extract method and the adaptive number threshold of inner point method are proposed. The purification results of mRANSAC are generally 60% to 300% more than those of RANSAC. Through setting suitable threshold value of mRANSAC, the purification rate can reach approximately 100%. This method can also be applied to solve the similar purification problem in other fields.
引用
收藏
页码：163 / 167
页数：4
相关论文
共 10 条
  • [1] Fischler M.A., Bolles R.C., Random sample consensus: A paradigm for model fitting with applications to image analysis and automated cartography, Communications of the Association for Computing Machinery, 24, 6, pp. 381-395, (1981)
  • [2] Lowe D.G., Distinctive image features from scale-invariant key points, International Journal of Computer Vision, 60, 2, pp. 91-110, (2004)
  • [3] Morel J.M., Yu G., ASIFT: A new framework for fully affine invariant image comparison, SIAM Journal on Imaging Sciences, 2, 2, pp. 438-469, (2009)
  • [4] He B., Zhu M., Improved fully affine invariant SIFT-based image matching algorithm, Optics and Precision Engineering, 19, 10, pp. 2472-2477, (2011)
  • [5] Matas J., Chum O., Randomized RANSAC with T<sub>d,d</sub> test, Image and Vision Computing, 22, 10, pp. 837-842, (2004)
  • [6] Chum O., Matas J., Optimal randomized RANSAC, Pattern Analysis and Machine Intelligence, 30, 8, pp. 1472-1482, (2008)
  • [7] Gong S., Zhao W., Liu C., Matched points purify algorithm based on gradient of disparity constrain, Journal of System Simulation, 20, S, pp. 407-410, (2008)
  • [8] Awwad T.M., Zhu Q., Du H., Et al., An improved segmentation approach for planar surfaces from unstructured 3D point clouds, The Photogrammetric Record, 25, 129, pp. 5-23, (2010)
  • [9] Gallo O., Manduchi R., Rafii A., CC-RANSAC: Fitting planes in the presence of multiple surfaces in range data, Pattern Recognition Letters, 32, 3, pp. 403-410, (2011)
  • [10] Xu M., Lu J., Distributed RANSAC for the robust estimation of three-dimensional reconstruction, IET Comput Vis, 4, 6, pp. 324-333, (2012)
1