Robust line segment matching via reweighted random walks on the homography graph

作者：

Highlights：

• Matching line segments for stereo images can be employed relying only on the geometry constraint.

• The association graph constructed with the epipolar constraint is invariant to the projective transformation.

• Employing the random walk on the association graph enables ranking the candidate.

• The proposed algorithm outperformed the state of the art methods in the experiments, especially in the scenes of the wide baseline, the steep viewpoint change, and the dense line segment.

摘要

•Matching line segments for stereo images can be employed relying only on the geometry constraint.•The association graph constructed with the epipolar constraint is invariant to the projective transformation.•Employing the random walk on the association graph enables ranking the candidate.•The proposed algorithm outperformed the state of the art methods in the experiments, especially in the scenes of the wide baseline, the steep viewpoint change, and the dense line segment.

论文关键词：Line segment matching,Epipolar geometry,Reweighted random walks,Graph matching

论文评审过程：Received 31 July 2019, Revised 29 September 2020, Accepted 7 October 2020, Available online 8 October 2020, Version of Record 19 October 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107693