Fourier-based multi-references shape prior for region-based active contours

In this paper we intend to present an original approach that incorporates geometric shape prior into region-based active contours. A Fourier-based shape alignment method is used to define prior knowledge from a reference shape. A set of complete and stable invariants to Euclidean transformations computed using Fourier transform on contours is used to choose the most suitable reference shape to the evolving contour. The proposed energy is then invariant with respect to rigid transformations. Experimental results show the ability of the geometric shape prior to constrain an evolving curve meet target shape and show the new method benefits on segmentation results in presence of occlusion, clutter and noise.