A novel photometric motion approach

Photometric motion is a computer vision process for the estimation of shape from image sequences of dynamic scenes. It was introduced by Pentland, based on his observation that the photometric effects of motion: the intensity change of a moving point, could dominate the purely geometric ones, due to projective distortion. We present a novel formulation for the same shape estimation problem, whose distinctive feature is that of being based on the irradiance change, due to the motion, at a given point in the image plane, and not, as in the original proposal, at a fixed location on the moving surface. Thus we obtain an easily implementable procedure which yields good-quality estimates, and which can be extended to single-input shape reconstruction.