DENSITY PROPAGATION FOR SURFACE TRACKING

This paper describes a novel approach to surface tracking in volumetri c image stacks. It draws on a statistical model of the uncertainties i nherent in the characterisation of feature contours to compute an evid ential field for putative inter-frame displacements. This field is com puted using Gaussian density kernels which are parameterised in terms of the variance-covariance matrices for contour displacement. The unde rlying variance model accommodates the effects of raw image noise on t he estimated surface normals. The evidential field effectively couples contour displacements to the intensity features on successive frames through a statistical process of contour tracking. Hard contours are e xtracted using a dictionary-based relaxation process. The method is ev aluated on both MRI data and simulated data. (C) 1998 Elsevier Science B.V. All rights reserved.