NONLINEAR MOTION ESTIMATION USING THE SUPERCOUPLING APPROACH

This paper presents the application of a very efficient multiresolutio n transformation, which is related to the renormalization group approa ch of physics, to the problem of motion segmentation. The proposed app roach is much faster and yields much better results than the full reso lution approach. The problem is formulated as one of global optimizati on where a cost function is constructed to combine the information obt ained by various processors as well as the constraints we impose to th e problem. The cost function is optimized using the supercoupling mult iresolution approach.