Last update: Sept. 7, 1998

Publications by Edwin R. Hancock et al. on Relaxation Labelling :

• Density Propagation for Surface Tracking, 1998.
• Feature Tracking by Multi-Frame Relaxation, 1995

BibTeX references.

Uses a statistical model of the uncertainties inherent in the characterization of feature contours to compute an evidential field for putative inter-frame displacements. The field is computed using Gaussian density kernels which are parametrized in terms of the variance & co-variance matrices for contour displacement. The underlying variance model accomodates the effects of raw image noise on the 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 extracted using a dictionary-based relaxation process.

An uncertainty ellipse is defined for the displacements on the frame L-1, in terms of the eigenvectors and eigenvalues of the related covariance matrix.

The computed uncertainties in the displacement components are used to construct probability density functions to model the contour evolution process. A Bayes formula is derived which couples feature characteristics derived from the filtering of frame L with the displacement field derived from hard-labelled contours on frame L-1.

Page created & maintained by Frederic Leymarie, 1998.
Comments, suggestions, etc., mail to: leymarie@lems.brown.edu