MULTI-CHANNEL COMBINED LOCAL GLOBAL OPTICAL FLOW APPROACHUSING PHOTOMETRIC INVARIANTS

A fundamental problem in the processing of image sequences is the measurement of optical flow (or image velocity). The goal is to compute an approximation to the 2-D motion field - which is the projection of the 3-D velocities of surface points onto the imaging surface - from spatiotemporal patterns of image intensity. Optical flow techniques can be classified into local approaches which are robust under noise, global approaches that yield dense flow fields, and finally into combined local global approaches that combine the advantages of both the local and the global approaches. Unfortunately, most the previous approaches assume that a point occurring in the scene has the same intensity in every frame. However, in real-life applications this is not the case where different illumination changes and shadows may occur. Therefore, Photometric invariants were proposed by Y.Mileva to solve this problem by applying them to a multi-channel global approach which resulted in a better robustness under illumination changes. Hence, we proposed to extend this work by applying photometric invariants to a multi-channel combined local global approach aiming to.formulate a model that is robust under noise and illumination changes and also provides dense flow fields. Experimental Results showed that the proposed method significantly performs better than the multi-channel global approach.