Most approaches to camera motion estimation from image sequences require matching the projections of at least 4 non-coplanar points in the scene. The case of points lying on a plane has only recently been addressed, using mainly projective cameras. We here study what can be recovered from two uncalibrated views of a planar contour under affine viewing conditions. We prove that the affine epipolar direction can be recovered provided camera motion is free of cyclorotation. The proposed method consists of two steps: 1) computing the affinity between two views by tracking a planar contour, and 2) recovering the epipolar direction by solving a second-order equation on the affinity parameters. Two sets of experiments were performed to evaluate the accuracy of the method. First, synthetic image streams were used to assess the sensitivity of the method to controlled changes in viewing conditions and to image noise. Then, the method was tested under more realistic conditions by using a robot arm to obtain calibrated image streams, which permit comparing our results to ground truth.

This work was supported by the project 'Perception, action & cognition through learning of object-action complexes.' (4915).

International Conference on Advanced Concepts for Intelligent Vision Systems (ACIVS) 2006, Antwerp (Belgium)

Peer Reviewed