基于不同尺度增强融合的视网膜血管分割算法

针对现有视网膜血管图像提取细小血管准确率较低的问题,提出了一种基于多尺度线性检测器与局部和全局增强相结合的视网膜血管分割方法.对多尺度线检测器进行研究,将其分为小尺度和大尺度两部分;利用小尺度对局部增强后的图像与大尺度对全局增强后的图像分别进行检测,得到不同尺度下的响应函数;将不同尺度下的响应函数进行融合,得到最终的视网膜血管结构.在STARE和DRIVE两个数据库上进行实验,结果表明:该算法得到的平均血管准确率分别达到96.62％和96.45％,平均真阳性率分别达到75.52％和83.07％,分割准确率高,能够得到较好的血管分割结果.

Segmentation algorithm for retinal vessels based on different scale enhancement and fusion

Aiming at problem of low accuracy of extracting the small blood vessels of existing retinal images,propose a novel retinal blood vessels segmentation method which is based on combination of multi-scale line detector and both local and global enhancement.The multi-scale line detector is divided into small scales and large scales;detect the local enhanced image with the small scales and the global enhanced image with the large scales,and get the response functions with different scales,respectively;fuse the response functions under different scales to get the final structure image of retinal blood vessels.In the STARE and DRIVE databases,the average blood vessels accuracy rates reach 96.62 ％ and 96.45 ％,respectively,and the average true positive rates reach 75.52 ％ and 83.07 ％,respectively.Segmentation accuracy is high,which can get better result of blood vessels segmentation.