A New Region-based Active Contour Model for Object Segmentation

We present a novel region based active contour model, that segments one or more regions of an input image that are visually similar to a given object of interest, said prior. The regions and the prior are described by the probability density function (pdf) of a visual feature, like color or intensity. Our active contour model deforms iteratively an initial region of the image making its pdf as close as possible to the pdf of the prior. This model is heuristic-based because the region evolution is driven by heuristic rules and it is not derived by minimizing an energy functional as in the classic variational approaches. The experiments carried out on real-world and synthetic datasets show very good performances, making it a valid alternative to the classic models.