【医学检测】基于计算机视觉实现视网膜血管检测附matlab代码

1 简介

针对眼底视网膜提取过程存在光照不均,噪声干扰等问题,采用包括光照均衡,图像增强,图像分割,二值化处理,最大类间方差法提取血管图像等预处理操作,解决图像光照不均,对象物呈球状分布,视网膜血管的分割难以达到理想分割效果等困难.提取包括深层次血管在内的完整眼底视网膜图像,达到较好效果.

2 部分代码

function [GreenChannel,AreaTrackingWhite,AreaTrackingGray,Vs,AreaTrackingWhite2,VsModif,dilateEdge] = FnTrackInit8(LS,Property) 
if(Property==1) % if the input is rgb image
  I=imresize(LS, 1);
  IG=I(:,:,2);  
elseif(Property==2) % if the input is  grayscale image
  IG=LS;
elseif(Property==3) % if the input is rgb at interval [0..1]
  IG=uint8(LS*255);
end
% preprosesing 
edgeRetina = edge(IG,'sobel',0.15);
% figure, imshow(edgeRetina);
se = strel('disk',5);
dilateEdge = imdilate(edgeRetina, se);
GreenChannel=IG;
% GreenChannel=adapthisteq(GreenChannel);
% IG=rgbImage;
[baris,kolom]=size(IG);
[counts,x] = imhist(IG,256);
bykPixel = baris*kolom;
% Initialization Tlow and Thigh base counting total pixel
TlowProsen = 0;
ThighProsen = 100;
jmlPixTlow = floor((TlowProsen/100)*bykPixel); % round don
jmlPixThigh = bykPixel - ceil((ThighProsen/100)*bykPixel); % round up
% count from front
jmlPix=0;
index1=1;
while(jmlPix<jmlPixTlow)
    jmlPix=jmlPix+counts(index1);
    index1=index1+1;
end
TlowI = x(index1);
% count from back
jmlPix=0;
index2=256;
while(jmlPix<jmlPixThigh)
    jmlPix=jmlPix+counts(index2);
    index2=index2-1;
end
ThighI = x(index2);
% Initialization matrix
IG_1=zeros(baris,kolom);
IG_2=IG;
IG_3=zeros(baris,kolom);
% subplot(2,4,4); imshow(IG_1); title('Chanel Vs (White)');
% Determine length of Vs
Vs=find(IG<=ThighI & IG>=TlowI); % Save index IG that satisfy find logic
elseVs1=find(IG<TlowI); % Save index IG that satisfy find logic
elseVs2=find(IG>ThighI); % Save index IG that satisfy find logic
IG_1(Vs)=IG_1(Vs)+1;
IG_2(elseVs1)=IG_2(elseVs1)*0; 
IG_2(elseVs2)=IG_2(elseVs2)*0;
AreaTrackingWhite=IG_1;
AreaTrackingGray=IG_2;
VsModif=Vs;
IG_3(VsModif)=IG_3(VsModif)+1;
AreaTrackingWhite2=IG_3;

3 仿真结果

4 参考文献

[1]师一帅. 基于分数阶微分的视网膜血管图像边缘检测[D]. 宁夏大学, 2013.

[2]张志顺. 眼底视网膜血管检测研究[J]. 现代计算机, 2015(23):10-13.

博主简介：擅长智能优化算法、神经网络预测、信号处理、元胞自动机、图像处理、路径规划、无人机等多种领域的Matlab仿真，相关matlab代码问题可私信交流。

部分理论引用网络文献，若有侵权联系博主删除。