SIMULTANEOUS MOTION ESTIMATION AND SEGMENTATION

We present a Bayesian framework that combines motion (optical Bow) est imation and segmentation based on a representation of the motion field as the sum of a parametric field and a residual field. The parameters describing the parametric component are found by a least squares proc edure given the best estimates of the motion and segmentation fields. The motion field is updated by estimating the minimum-norm residual fi eld given the best estimate of the parametric field, under the constra int that motion field be smooth within each segment. The segmentation field is updated to yield thf minimum-norm residual field given the be st estimate of the motion field, using Gibbsian priors. The solution t o successive optimization problems are obtained using the highest conf idence first (HCF) or iterated conditional mode (ICM) optimization met hods. Experimental results on real video are shown.