Three methods of detecting road edges in millimeterwaue radar images are presented i n this paper. All of t h e m are based o n deformable template priors and rand o m field likelihoods. T h e first method is formulated in a Bayesian setting and employs a n adaptive M A P esti- mate. T h e second method is a modification of the first, using a novel weighting scheme. T h e third method is based o n a three-region indicator matrix which is used to impose the non-linear constraints implicit o n

more »

... d geometry via addition of a s u m of quasi-quadratic matrix f o r m s to the log-normal likelihood. Unlike the first two methods, that employ the Metropolis algorithm t o find the optimal road edges, the third method uses a deterministic recursive scheme designed to find the optimal indicator matrix. Experimental results are presented to show the advantages of these methods.