|
Abstract : |
Abstract. We interpret a classical method for optical flow computation due to Nagel and Enkelmann (1986) as a tensor-driven anisotropic diffusion approach in digital image analysis. We introduce an improvement into the model formulation, and we establish well-posedness results for the resulting system of parabolic partial differential equations. Since our method avoids linearizations in the optical flow constraint, it can recover displacement fields which are far beyond the typical one-pixel limits that are characteristic for differential methods for optical flow recovery. A robust numerical scheme is presented in detail. We avoid convergence to irrelevant local minima by embedding our method into a linear scalespace framework and using a focusing strategy from coarse to fine scales. We illustrate the high accuracy of the proposed method by means of a synthetic and a real-world image sequence. 1, |
