3D Object Reconstruction using Line and Point Feature Matching

S. Chang, H. Deng, J. Fuller, and A. Farsaie (USA)


Stereo vision, 3D line reconstruction, line-to-point matching, point-to-point matching, point features correlation, line features correlation


In this paper we present a three-stage matching algorithm for 3D line-based model reconstruction from stereo image pairs. In the first stage, a line-to-point (LTP) matching is used to determine if a line feature in the first image is related to a point feature in that same image. In the second stage, the point features detected in the first stage are used to match feature points in the second image by a point-to-point (PTP) matching. In the third stage, the LTP matching is used again to find related line features that are associated with the corresponding points in the second image. PTP and LTP establish the correlation between line features in the first and second images. This algorithm combines the advantages of point feature matching as well as line segment matching.

Important Links:

Go Back