查询结果:   于挥,王小鹏.基于HESSIAN增强和形态学尺度空间的视网膜血管分割[J].计算机应用与软件,2016,33(8):200 - 205.
中文标题
基于HESSIAN增强和形态学尺度空间的视网膜血管分割
发表栏目
图像处理与应用
摘要点击数
399
英文标题
RETINAL VESSELS SEGMENTATION BASED ON HESSIAN ENHANCEMENT AND MORPHOLOGICAL SCALE SPACE
作 者
于挥 王小鹏 Yu Hui Wang Xiaopeng
作者单位
兰州交通大学电子与信息工程学院 甘肃 兰州 730070     
英文单位
School of Electronics and Information Engineering,Lanzhou Jiaotong University,Lanzhou 730070,Gansu,China     
关键词
视网膜血管 Hessian增强 尺度空间 形态学分割
Keywords
Retinal vessel Hessian enhancement Scale space Morphological segmentation
基金项目
国家自然科学基金项目(61261029);兰州市科技计划项目(2013-4-63);金川公司预研基金项目(JCYY20130 09)
作者资料
于挥,硕士生,主研领域:图像处理。王小鹏,教授。 。
文章摘要
眼底视网膜血管的走向、弯曲度、分叉度等性状分析已成为医学上诊断全身血管性疾病的重要手段。采集到的眼底图像常存在光照不均匀等现象,利用传统的血管分割方法难以对微小血管进行检测。为此提出一种基于改进Hessian矩阵增强和形态学尺度空间的分割方法。首先利用高斯函数构建多尺度Hessian增强滤波器,采用新型的血管相似性函数对血管网络进行对比度增强,同时平滑图像以减轻噪声;然后利用改进的Top-hat变换尺度空间从背景中提取血管,并引入形态学重建方法进一步突出血管像素,消除伪边缘及孤立点噪声;最后使用二次阈值化方法实现血管的最终分割。仿真结果表明,改进的分割方法在保证大血管脉络准确分割的同时,能够较好地实现微小血管分割。
Abstract
Characters analysis in regard to the trend, curvature and bifurcation of retinal vessels in fundus has become the important means of systemic vascular diseases diagnosis in medicine science. Because of most collected fundus images has the phenomenon of light unevenness, it is difficult to use traditional vessel segmentation methods to detect the micro vessels. Therefore we proposed a segmentation algorithm, it is based on the improved Hessian matrix enhancement and morphological scale space. First, by using Gauss function the algorithm constructs multi-scale Hessian enhanced filter, and uses a novel vascular similarity function to carry out the contrast enhancement on vascular network, while smoothes the image to weaken noise as well; then it extracts the vessels from background using an improved Top-hat transformation scale space, and introduces morphological reconstruction method to further highlight the vascular pixels and to eliminate the pseudo-edges and the noise of outliers; finally the algorithm uses secondary thresholding approach to realise final vessel segmentation. Simulation experimental results showed that while ensuring the accurate segmentation of great vessels and choroid, the improved segmentation method can better realise the segmentation of micro vessels.
下载PDF全文