A Two-Stage Algorithm for Motion Discontinuity-Preserving Optical Flow Estimation

continuity constraint of motion vectors. Also, many reThis paper proposes a two-stage optical flow estimation algosearchers (e.g., Horn and Schunck [10], Yachida [11], rithm for video sequences based on feature-based matching and Glazer [12], Nagel [13], Nagel and Enkelmann [14], Paquin module-fusion. In the first stage, the algorithm computes initial and Dubois [15], Hildreth [16], Mitiche et al. [17], Aisbett translation and rotation parameters using a relaxation scheme [18], Schnörr [19], and Black and Anandan [20]) have proat feature points and object-oriented anisotropic propagation posed gradient-based optical flow estimation methods. in continuous and discontinuous regions. In the second stage, These approaches made use of a motion constraint equaiterative-improvement optical flow estimation based on the tion and some other constraints. Usually the optical flow strongly coupled recurrent module-fusion and adaptive field over an entire image can be calculated by using the smoothing is introduced. Computer simulation results of the smoothness constraint. These approaches mostly exproposed and conventional algorithms are shown.  1997 Acapressed the smoothness constraint in terms of a partial demic Press