Local Statistical Active Contour Energy Functional based on Cauchy-Schwarz Divergence for Image Segmentation

Abstract

Active contour model is a widely used method for image segmentation. However, traditional active contour models often fail to segment objects in images with intensity inhomogeneity. In this paper, a local statistical active contour energy functional is proposed for image segmentation. It employs Cauchy-Schwarz divergence as the statistical distance measuring the difference between the estimated probability density inside and outside the evolving contour, which extracts local intensity information to guide the evolution of the contour. The devised energy functional can well integrate with traditional active contour model, thus it can help achieve improvement for traditional active contour models when they are used for segmenting objects out of intensity inhomogeneous backgrounds. Experimental results on synthetic and real images show that the proposed local energy functional improves the performance of traditional active contour models in images with intensity inhomogeneity.