An active contour model for segmentation based on cubic B-Splines and gradient vector flow

Abstract

The aim of this paper is to present advances in segmentation for visualization and quantitative analysis in bioimaging. Here, we combine two existing approaches for segmentation with snakes. Firstly, we use cubic B-splines to represent the snake using coarse-to-fine control point insertion; this allows to smooth adaptively the resulting contour while reducing the risk to get attracted from misdetected edges. Secondly, we put the snake in a gradient vector flow (GVF) field. This enables the snake to evolve into concavities of the shape. Further, sensitive parameters drop out in our setting and the attraction range with respect to initialization of the snake is enlarged.