[image segmentation] blood vessel extraction from retinal fundus images based on maximum principal curvature with matlab code

1 brief introduction ​

In order to reduce the number of missed and false detection of low contrast vessels in the process of retinal vascular skeleton extraction , This paper presents a method of extracting retinal vascular skeleton based on principal curvature . First , Extract the principal curvature of each pixel in the retinal image after multi-scale Gaussian filtering ; secondly , The local maximum points in the direction of maximum principal curvature are extracted at each scale , The high contrast blood vessel center pixels are selected as seed points by curvature threshold ; Skeleton tracking and labeling of low contrast vessels ; Last , Fusion of vascular skeletons extracted at multiple scales .

2 Part of the code

function [lamdaplus,lamdaminus]=lamdafind(gxx1,gyy1,gxy1)%This function perform finding eigen values of hessian matrix and output%maximum and minimum eigen value as lamdaplus and lambaminus%gxx1: Second derivative over x%gyy1: Second deriative over y%gxy : derivative over x and y %Author : Achintha Iroshan ,University of Moratuwa​%Hessian matrix -> H = [gxx gxy;gyx gyy] .Since hessian matrix is %symetrical with real eigen values gxy = gyx​H=[gxx1 gxy1;gxy1 gyy1];​%Obtain eigen valueslamda=eig(H);​%Obtain maximum and minimum lamda valuesif lamda(1)>lamda(2)    lamdaplus = lamda(1);    lamdaminus = lamda(2);else if lamda(1)<lamda(2)    lamdaplus = lamda(2);    lamdaminus = lamda(1);      else        lamdaplus = lamda(1);        lamdaminus = lamda(2);    endendend

3 Simulation results

4 reference

[1] Yuanweiqi , Wang an . A multi-scale retinal vascular skeleton extraction method based on principal curvature and principal direction [J]. Journal of instrumentation , 2021.

About bloggers ： Good at intelligent optimization algorithms 、 Neural networks predict 、 signal processing 、 Cellular automata 、 The image processing 、 Path planning 、 UAV and other fields Matlab Simulation , relevant matlab Code problems can be exchanged by private letter .

Some theories cite network literature , If there is infringement, contact the blogger to delete .