改进主动轮廓线模型在图像分割中的应用

摘要

ACM (Active Contour Model), also called Snake model, is a usual image segmentation algorithm. In ACM based image segmentation, it is difficult to segment concavity edge and weak boundary region. In this paper, a local profile model is combined with original ACM model. It can confirm the real edge through comparing local profile and the pixels along the normal line of the key points. A new method to compute the curvature energy of the key points is also introduced which can fit the ACM model to concavity edge and improve the convergence speed. Experimental results validate the performance of improved ACM model.%主动轮廓线模型(Active Contour Model,ACM),也称作蛇(Snake)模型,是一种常用的图像分割算法.在基于主动轮廓线的图像分割中,深度凹陷边界的逼近和弱边界区域的分割一直是一个难点.引入了一种局部纹理模型(Local Profile Model)匹配算法,通过匹配沿控制点法线方向像素和局部纹理模型可以确定弱边界区域的真实边界,并结合一种新的计算控制点曲率外力的算法,使得主动轮廓线模型能够逼近图像的深度凹陷区域的同时提高算法的收敛速度.实验结果表明,该方法是有效的.