Study On Optical Flow Computation Methods For Video Image

The apparent pixel motion in an image sequence, called optical flow, is a useful primitive for dynamic scene analysis. Many domestic and overseas scholars insist on the investigation of optical flow algorithms. Many algorithms such as differential optical flow algorithm, matching optical flow algorithm, energy-based optical flow algorithm and phase-based optical flow algorithm are proposed. However, the computation of flow velocity is an ill-posed problem. Furthermore, with the increase of velocity, and the gray discontinuity, the error of brightness constant assumption will increase, which reduces the computation accuracy. Meanwhile, the computation of optical flow need iterative, so the computation rate is slow, which reduces the application.Firstly, the basic principle of optical flow calculation is analyzed, and the key techniques are discussed in the dissertation, which the gradient-based optical flow algorithm is focused on. Secondly, a new method to calculate optical flow based on gradient aim at brightness discontinuities is presented. This method adds counter restraint at brightness discontinuities, so the robustness of the method is increased. Finally, a multigrid-based optical flow algorithm is presented against the problem of the slow computation. Simulating experiments show that this method can reduce the iteratives, and improve the calculate rate.