摘要
视网膜血管自动分割能辅助诊断某些眼底疾病和系统性血管疾病。为了提高血管自动分割的效率,因此提出了一种线算子引导Gabor小波的视网膜血管分割方法。利用线算子检测血管方向的最优匹配角,将其作为Gabor小波变换的旋转角构建4个不同尺度的Gabor小波,并提取4维Gabor小波特征,加上两个线强度和预处理后的图像灰度,构建7维特征向量,采用SVM进行分类。与其他基于Gabor小波的方法相比,本方法只需计算最优匹配角所对应方向的Gabor小波特征,大大降低了多尺度Gabor小波特征提取的计算量,此外线算子特征与Gabor小波特征的良好互补性,有利于提高血管与背景的辨别度。在DRIVE眼底数据库上进行实验,其平均准确率、灵敏度及特异性分别为0.9361、0.8238及0.9554,获得了不错的分割性能。
Automated retinal vessel segmentation can aid in the diagnosis of certain fundus diseases and systemic vessel diseases.In order to improve the efficiency of automatic segmentation of blood vessels.Therefore,a retinal vessel segmentation method based on Gabor wavelet guided by line operator is proposed.The line operator is used to detect the optimal matching angle of the vessel direction,and the optimal matching angle is used as the rotation angle of the Gabor wavelet transform to construct four different scale Gabor filters,and the 4-D Gabor wavelet feature is extracted.The features of two line operators and the pre-processed image gray value are added to construct a 7-D feature vector,which are classified by SVM.Compared with other methods based on Gabor wavelet,this paper only needs to compute the Gabor wavelet features in the direction corresponding to the optimal matching angle,which greatly reduces the computational complexity of multi-scale Gabor wavelet feature extraction.In addition,the good complementarity between the line operator feature and the Gabor wavelet feature is beneficial to improve the blood vessel and the degree of discrimination of the background.Experiments are carried out on the fundus image database DRIVE.The average accuracy,sensitivity,and specificity of the method in this paper are 0.9361,0.8238 and 0.9554,respectively.and good segmentation performance is obtained.
作者
李灿标
郑楚君
LI Canbiao;ZHENG Chujun(School of Physics and Telecommunications Engineering,South China Normal University,Giuingzhou 5\006 yChina)
出处
《激光杂志》
北大核心
2020年第1期185-191,共7页
Laser Journal
基金
国家自然科学基金(No.10504008)
