Statistical neighbor distance influence in active contours

Abstract

In this paper, we propose a new model for segmentation of images containing multiple objects. In order to take advantage of the constraining information provided by neighboring objects, we incorporate information about the relative position and shape of neighbors into the segmentation process by defining a new “distance” term into the energy functional. We introduce a representation for relative neighbor distances, and define a probability distribution over the variances of the relative neighbor distances of a set of training images. By minimizing the energy functional, we formulate the model in terms of level set functions, and compute the associated Euler-Lagrange equations. The contours evolve both according to the relative distance information and the image grey level information. Several objects in an image can be automatically detected simultaneously.