Abstract
Curve evolution schemes are introduced for image segmentation based on a region-based contour model allowing for junctions, vector-valued images, and topology changes. Together with an a posteriori denoising in the segmented homogeneous regions, this leads to a fast and efficient method for image segmentation and restoration. An uneven spread of mesh points is avoided by using the tangential degrees of freedom. Several numerical simulations on artificial test problems and on real images illustrate the performance of the method.
Full Text
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call