ABSTRACT
This chapter focuses on the optical flow technique and addresses some fundamental issues associated with optical flow. It discusses the differential method and presents a multiple attributes approach. The chapter explores some performance comparisons between various techniques. It also discusses multiple attribute techniques in optical flow determination. Under the assumption of the symmetric response distribution with a single maximum value assumed by the ground-truth optical flow, the convergence of the correlation-feedback technique is justified. Except if the optical flow vectors happen to have only an integer multiple of pixels as their components, an analysis in shows that the correlation-based approach will not converge to the apparent two-dimensional (2-D) motion vectors and will easily have error much greater than 10%. The optical flow field is a dense 2-D distribution of apparent velocities of movement of intensity patterns in image planes, while the 2-D motion field can be understood as the perspective projection of three-dimensional motion in the scene onto image planes.
