OPTICAL-FLOW BY NONLINEAR RELAXATION

In this paper, we present a local approach to the computation of raw o ptical now fields based on nonlinear relaxation, which is explicitly d esigned to be coupled with a nonlinear vector filtering technique so a s to achieve an effective rejection of the outliers. We derive the app roach from a standard linear algorithm-the one-dimensional least squar es. This is split down into two steps. First, the linear algorithm is formulated as an iterative procedure, in which image points in a suita ble geometric neighborhood are made to interact each other. Second, a nonlinearity term in the elementary interaction between neighbors is a dded, which depends on the Gaussian interaction of elements of a veloc ity neighborhood. The results of a comparison with two linear local te chniques show clearly the effectiveness of the approach, both in terms of motion boundary preservation and of optical flow field density.