3D Curves Reconstruction from Multiple Images

Abstract

In this paper, we propose a new approach for reconstructing 3D curves from a sequence of 2D images taken by uncalibrated cameras. A curve in 3D space is represented by a sequence of 3D points sampled along the curve, and the 3D points are reconstructed by minimizing the distances from their projections to the measured 2D curves on different images (i.e., 2D curve reprojection error). The minimization problem is solved by an iterative algorithm which is guaranteed to converge to a (local) minimum of the 2D reprojection error. Without requiring calibrated cameras or additional point features, our method can reconstruct multiple 3D curves simultaneously from multiple images and it readily handles images with missing and/or partially occluded curves.