ACTIVE CONTOURS DRIVEN BY LOCAL AND GLOBAL IMAGE FITTING ENERGY

摘要

An improved region-based active contour model in a variational level set formulation is proposed in this paper.We define an energy functional with a local intensity fitting term,which extract the local image information,and a global intensity fitting term,which drives the contour far away from object boundaries.In local intensity fitting term,geodesic time is defined as kernel function which provides an adaptive selection of field sampling pixels using spatial distance and image gradient.Therefore,the proposed model remains edge information while using larger bandwidth,and applied to images segmentation with intensity inhomogeneity.Experimental results show the advantages of our method in terms of accuracy and robustness.