A Robust Feature Points Matching Algorithm in 3D Optical Measuring System

Abstract:

In optical measuring system with a handheld digital camera, image points matching is very important for 3-dimensional(3D) reconstruction. The traditional matching algorithms are usually based on epipolar geometry or multi-base lines. Mistaken matching points can not be eliminated by epipolar geometry and many matching points will be lost by multi-base lines. In this paper, a robust algorithm is presented to eliminate mistaken matching feature points in the process of 3D reconstruction from multiple images. The algorithm include three steps: (1) pre-matching the feature points using constraints of epipolar geometry and image topological structure firstly; (2) eliminating the mistaken matching points by the principle of triangulation in multi-images; (3) refining camera external parameters by bundle adjustment. After the external parameters of every image refined, repeat step (1) to step (3) until all the feature points been matched. Comparative experiments with real image data have shown that mistaken matching feature points can be effectively eliminated, and nearly no matching points have been lost, which have a better performance than traditonal matching algorithms do.

• Periodical: Advanced Materials Research (Volumes 383-390)
• Pages: 5193-5199
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.383-390.5193
• Online since: November 2011
• Authors: Jian Ying Yuan, Xian Yong Liu, Zhi Qiang Qiu
• Keywords: 3D Reconstruction, Eliminating Erroneous Matching Points, Feature Matching, Multiple Images

References

[1] Lu Chengjing, Huang guiping, Li guangyun. Accuracy testing and application of V-STARS, Infrared and Laser engineering, 36, pp.245-249, 2007(in chinese).

[2] Zhou shikan, Lou zhenliang, Shu shixiang. A point cloud collecting method based on Atos & Tritop, Die and mould technology, vol. 2, pp.51-55, 2003(in chinese).

[3] Marc pollefeys. Visual modeling with a hand-held camera, International journal of computer vision, vol. 59, pp.207-232, Sep-Oct (2004).

[4] WANG Bao-feng FAN Sheng-hong HUANG Gui-ping. Matching Homologous Points Based on Photographs in Computer Vision[J]. Journal of zhengzhou institute of surveying and mapping , 2006, 23（6）; 451-453(in chinese).

[5] Zhang Zuxun, Yang shengchun, Zhang jianqing. Multi-baseline digital close-range photogrammetry[J]. Geospatial Information, 2007, 5(1): 1-4(in chinese).

[6] Zheng Xiaojie. Research of key technologies in close range photogrammetry applied in reverse engineering[D]. Shanghai: Shanghai Jiao Tong University, 2007; 33-41.

[7] Liang Dong, Tong Qiang, QuLei. Algorithm for images matching based on epipolar and homography constraints[J]. Journal of system simulation, 2006, 18（1）; 44-46(in chinese).

[8] CHENJ, CLARKE, TA, ROBSONS. An alternative to the epi-polar line method for automatic target matching in multiple ima-ges for 3D measurement[J] . Optical 3D Measurement Tech-niques II. 1994, 2252 : 197-204.

DOI: 10.1117/12.169837

[9] Richard Hartley, Andrew Zisserman. Multiple View Geometry in Computer Vision, Cambridge University Press, second edition, pp.239-240, (2003).

[10] Zhang qixian. The distance problem of two skew lines in space. Journal of longyan teachers college, vol. 20, pp.86-87, Dec. 2002(in chinese).

[11] Bill Triggs, Philip McLauchlan, Richard Hartley. Bundle adjustment-a modern synthesis[J], (2000).

[12] Marc Pollefeys, Reinhard Koch, Luc van gool. Self-calibration and metric reconstruction in spite of varying and unknown intrinsic camera parameters. International Journal of computer vision, (1998).

DOI: 10.1109/iccv.1998.710705

[13] Juyang Weng. Camera calibration with distortion models and accuracy evaluation, Transaction on pattern analysis and machine intelligence, vol. 14, pp.965-980, Oct, (1992).

DOI: 10.1109/34.159901

[14] Sung Joon Ahn, Wolfgang Rauh. Circular coded target for automation of optical 3D-measurement and camera calibration, International Journal of Pattern Recognition and Artificial Intelligence, vol. 15, pp.905-919, (2001).

DOI: 10.1142/s0218001401001222