演化曲线自适应驱动的图像分割水平集模型研究

A level-set model for image segmentation with adaptively driven evolutional curve

利用图像梯度和几何曲率等信息可以准确定位分割图像的边缘.基于此,本文在对图像分割典型变分模型有效性及所存在问题分析和讨论的基础上,提出了一种演化曲线自适应驱动的图像分割水平集模型.模型通过调整演化曲线长度项和面积项的权重函数,使演化曲线能够根据图像当前的状态自适应的调整演化幅度和方向,不仅提高了图像分割的准确度,还大大缩减了图像分割时间;模型在利用图像局部区域信息的同时,也利用全局化的正则函数来兼顾模型能量泛函的全局性,使模型有了对异质区域边界的捕捉能力.经试验验证文章所提出的新模型有效可靠.

Image segmentation model based on partial differential equation is widely concerned,because it can accurately locate the edge of object by directly using images geometric information such as gradient,curvature,and etc.In this study,we first discuss the effectiveness and disadvantages of state-of-the-art variational models for image segmentation.Based on our analysis,we subsequently propose a novel level-set model for image segmentation with adaptively driven evolutional curve.The features of the proposed model are as follows.According to current state of the image,the evolutional curve is able to adaptively adjust evolution amplitude and direction by introducing a weighting function of length and area terms.This both improves segmentation accuracy and reduces time to obtain ideal segmentation results,While depending on the local image information,aglobal regular function is employed to balance the global feature of energy functional so as to enhance the models ability to capture edges of heterogeneous area.Experimental results verify the effectiveness of the proposed model.