Recovering structure and 3-D motion parameters of a rigid planar patch using moments

It is shown that a binocular observer can recover the 3-D structure and motion of a rigid planar patch. This is accomplished by combining motion and stereo to obtain linear equations in the unknowns, which yield unique solutions. The approach is based on the aggregate stereo method and on the general theory of moments. The computation of a limited set of moments of four images is required. The images of the rigid planar patch involved are two before and two after the motion. Real-time application of the approach is possible, since image moments can be computed in real time (frame rates). Knowledge of point-to-point correspondences between the different images is not required. The approach presented solves the problem of motion estimation by looking at the object as an array of 3-D points. It is assumed that there is only one object in the image and that the object is undergoing rigid motion