Multiple light source optical flow

Abstract

A novel method is described to compute a dense, local representation of optical flow. The idea is to use the intensity values recorded from multiple images of moving objects acquired simultaneously under different conditions of illumination. Each image is assumed to satisfy the standard optical flow constraint equation. Multiple images give rise to multiple constraint equations. When the optical flow and the 2-D motion field coincide, these multiple equations are in the same unknowns. A description is given of the basic theory, and the theory is illustrated on a real image motion sequence. All computations are local, independent and relatively simple. No iteration steps are required. It is suggested that the requirement to obtain simultaneous images under different conditions of illumination be satisfied by using spectrally distinct illumination and sensing. >