Pairs of images can be brought into alignment (registered) by finding corresponding points on the two images and deform ing one of them so that the points match. This can be carried out as a Hamiltonian boundary-value pr oblem\, and then provides a diffeomorphic registra tion between images. However\, small changes in th e positions of the landmarks can produce large cha nges in the resulting diffeomorphism. We formulate a Langevin equation for looking at small random p erturbations of this registration. The Langevin eq uation and three computationally convenient approx imations are introduced and used as prior distribu tions. A Bayesian framework is then used to comput e a posterior distribution for the registration\, and also to formulate an average of multiple sets of landmarks. \; LOCATION:Seminar Room 1\, Newton Institute CONTACT:INI IT END:VEVENT END:VCALENDAR