Developing rigid constraint for the estimation of pose and structure from a single image

Pose and structure estimation from a single image is a fundamental problem in machine vision and in multiple sensor fusion and integration. While existing calibration algorithms are mainly based on invariants described in a single coordinate frame, in this paper we propose using rigid constraints described in different coordinate frames to iteratively estimate structural and camera pose parameters. The method is based on a coarse-to-fine strategy in which initial estimation is obtained through a classical linear algorithm. For a comparative study of performance, we also implemented an extended motion estimation algorithm (extended from the 2D-2D to the 3D-2D case) based on epipolar geometry. Experimental results based on both synthetic data and real images show that the iterative algorithm is more accurate and robust than both the extended algorithm and the classical linear algorithm.