Color optical flow estimation based on vector homogeneity analysis

A novel color video optical flow estimation approach based on vector homogeneity analysis is proposed, in which the optical flow equation (OPE) is established by the color constant model (CCM) rather than the brightness constant model (BCM) which is used in ordinary optical flow estimation approaches. In order to solve concisely, the CCM-based OPE is projected onto the homogeneity direction of a local color vector, and a homogeneity CCM (HCCM) OPE is obtained. The HCCM-based OPE can be solved with a BCM-similarly method. Because it contains not only luminance information but also chrominance information, the HCCM-based OPE can capture motions in color video more accurate than that of the BCM-based method. Finally, two experiments are given to show the proposed approachpsilas effectiveness.