Analysis of differential and matching methods for optical flow

A number of algorithms for optical flow are studied on both a theoretical and an experimental ground. Differential and matching methods are examined. Both types of optical-flow algorithms can use either local or global constraints, such as spatial smoothness, in computing the flow. Local matching and differential techniques and global differential techniques are examined. The traditional algorithms for optical flow utilize weak assumptions on the local variation of the flow, and on the variation of image brightness. Strengthening these assumptions improves the flow computation. The computational consequence of this is a necessity for larger spatial and temporal support. Using larger support is valid when constraints on the local shape of the flow are satisfied. Experiments show the behavior of these optical-flow methods on velocity fields which do not obey the assumptions. Implementation of these methods highlights the importance of numerical differentiation