On the Evaluation of Optical Flow Algorithms

Evaluating the performance of optical flow algorithms has been difficult because of the lack of ground-truth data sets for complex scenes. We describe a simple modification to a ray tracer that allows us to generate ground-truth motion fields for scenes of arbitrary complexity. The resulting flow maps are used to assist in the comparison of eight optical flow algorithms using two complex, synthetic scenes. We introduce a new performance metric based on the correlation of vector gradient magnitude images which is especially pertinent for evaluating optical flow algorithms for boundary detection applications.