Active contour model driven by Globally Signed Region Pressure Force

Abstract

One of the most popular and widely used global active contour models (ACM) is the region-based ACM, which relies on the assumption of homogeneous intensity in the regions of interest. As a result, most often than not, when images violate this assumption the performance of this method is limited. Thus, handling images that contain foreground objects characterized by multiple intensity classes present a challenge. In this paper, we propose a novel active contour model based on a new Signed Pressure Force (SPF) function which we term Globally Signed Region Pressure Force (GSRPF). It is designed to incorporate, in a global fashion, the skewness of the intensity distribution of the region of interest (ROI). It can accurately modulate the signs of the pressure force inside and outside the contour, it can handle images with multiple intensity classes in the foreground, it is robust to additive noise, and offers high efficiency and rapid convergence. The proposed GSRPF is robust to contour initialization and has the ability to stop the curve evolution close to even ill-defined (weak) edges. Our model provides a parameter-free environment to allow minimum user intervention, and offers both local and global segmentation properties. Experimental results on several synthetic and real images demonstrate the high accuracy of the segmentation results in comparison to other methods adopted from the literature.