Recognition Using Motion And Shape

Our goal is to design a recognition system which can distinguish between two objects with the same shape, but different motion or between two objects with the same motion but a different shape. In this paper, we present a method for matching sets of trajectories which supplements motion information with knowledge about the spatial relationships between points on the moving object. First we present a simple algorithm which matches two single trajectories using only motion information. We convert the 2-D motion trajectories into two 1-D signals based on the speed and direction components. The signals are then represented by scale-space images both to simplify matching and because the scale-space representations are translation and rotation invariant. We extend the matching algorithm to include spatial information and propose a second algorithm which matches multiple trajectories by combining motion and spatial match scores. Both algorithms were tested with real and synthetic data