Computation Optical Flow Using Pipeline Architecture

Abstract

Accurate estimation of motion from time-varying imagery has been a popular problem in vision studies, This information can be used in segmentation, 3D motion and shape recovery, target tracking, and other problems in scene analysis and interpretation. We have presented a dynamic image model for estimating image motion from image sequences, and have shown how the solution can be obtained from a set of partial differential equations. In this paper, we have investigated a relaxation type algorithm for obtaining a numerical solution to these equations, and considered the implementation of the algorithm on a variation of the general pipeline interconnection scheme using transputers. This architecture is compared against two others based on flexibility and efficiency. It is observed that with respect to computation, a mesh connected architecture has advantages over the proposed pipeline scheme. However, the pipeline configuration is easily expandable and more robust to changes in the algorithms parameters and image size.