基于局部和全局活动轮廓的医学图像分割模型  

Medical image segmentation model based on local and global active contour

作　　者：李佳宇 陈利霞[1] LI Jiayu;CHEN Lixia(School of Mathematics and Computational Science,Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学数学与计算科学学院

出　　处：《桂林电子科技大学学报》2019年第5期369-373,共5页Journal of Guilin University of Electronic Technology

基　　金：国家自然科学基金(61762025,61661017);广西自然科学基金(2018GXNSFAA138169);桂林电子科技大学研究生教育创新计划(2017YJCX84);广西高等教育本科教学改革工程项目(2017JGB230);广西高校数据分析与计算重点实验室项目(LDAC201704);广西重点研究发展计划(AB17195053)

摘　　要：针对医学图像容易受到各种因素干扰而出现灰度不均匀和边界模糊,从而增加其分割难度的问题,提出一种基于局部和全局活动轮廓的医学图像分割模型。该模型先引入一组正交基函数重新定义局部拟合图像,保证偏置域缓慢平稳变化;再利用k近邻分类算法得到概率分数,构造新的加权能量函数。实验结果表明,相较于对比模型,本模型对图像灰度不均匀和边界模糊图像的分割效果更好,对于噪声更加鲁棒。Medical images are easily interfered by various factors,resulting in intensity inhomogeneity and blurred boundaries,which increases the difficulty of segmentation.Aiming at these problem,a medical image segmentation model based on local and global active contour is proposed.Firstly,a set of orthogonal basis functions are introduced into the new model to redefine the local fitted image,so as to ensure a slow and steady change in the bias domain;secondly,the k-nearest neighbor classification algorithm is utilized to obtain the probability score,which is used to construct a new weighted energy function.Experimental results show that,compared with the contrastive models,the proposed model has better segmentation effect on image intensity inhomogeneity and fuzzy boundary problem,and has better robust to noise.

关 键 词：图像分割 医学图像 水平集 机器学习 基函数 概率得分 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]