【图像分割】基于电磁算法优化多级阈值实现图像分割附matlab代码

原创

matlab科研助手 2022-10-17 22:17:13 ©著作权

©著作权归作者所有：来自51CTO博客作者matlab科研助手的原创作品，请联系作者获取转载授权，否则将追究法律责任

✅作者简介：热爱科研的Matlab仿真开发者，修心和技术同步精进，matlab项目合作可私信。
🍎个人主页：Matlab科研工作室
🍊个人信条：格物致知。

⛄ 内容介绍

针对传统Kapur熵和oust在多阈值图像分割算法中存在运算量大、计算效率低以及精度不高等问题,提出了一种基于电磁算法的多级阈值图像分割方法,该方法采用Kapur熵作为计算适应度的目标函数,通过引入电磁算法求解目标函数最大化时的全局最优问题.实验结果表明:相对于其他方法,本文方法在多个评价指标上都有很好的性能体现,并且本文方法在保证较好分割效果的同时,计算效率明显提升.

⛄ 部分代码

%Diego Oliva, Erik Cuevas, Gonzalo Pajares, Daniel Zaldivar, Valent韓 Osuna.

%A Multilevel Thresholding algorithm using electromagnetism optimization

%Universidad Complutense de Madrid / Universidad de Guadalajara

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%The algorithm was published as:

%Diego Oliva, Erik Cuevas, Gonzalo Pajares, Daniel Zaldivar, Valent韓 Osuna.

%A Multilevel Thresholding algorithm using electromagnetism optimization,

%Journal of Neurocomputing, 139, (2014), 357-381.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [fitR, fitBestR, fitG, fitBestG, fitB, fitBestB] = fitnessIMG(I, N, Lmax, level, xR, probR, xG, probG, xB, probB)

%Metodo de Otsu

%Evalua poblaciones xR, xG, xB, en la funcion objetivo para obtener

%fitR, fitG, fitB, dependiendo si la imagen es RGB o escala de grises

for j = 1:N

if size(I,3) == 1

%grayscale image

fitR(j) = sum(probR(1:xR(j,1))) * (sum((1:xR(j,1)) .* probR(1:xR(j,1)) / sum(probR(1:xR(j,1)))) - sum((1:Lmax) .* probR(1:Lmax)) ) ^ 2;

for jlevel = 2:level - 1

fitR(j) = fitR(j) + sum(probR(xR(j,jlevel - 1) + 1:xR(j,jlevel))) * (sum((xR(j,jlevel - 1) + 1:xR(j,jlevel)) .* probR(xR(j,jlevel - 1) + 1:xR(j,jlevel)) / sum(probR(xR(j,jlevel - 1) + 1:xR(j,jlevel)))) - sum((1:Lmax) .* probR(1:Lmax))) ^ 2;

end

fitR(j) = fitR(j) + sum(probR(xR(j,level-1) + 1:Lmax)) * (sum((xR(j,level - 1) + 1:Lmax) .* probR(xR(j,level - 1) + 1:Lmax) / sum(probR(xR(j,level - 1) + 1:Lmax))) - sum((1:Lmax) .* probR(1:Lmax))) ^ 2;

% if isnan(fitR(j))

% fitR(j)=eps;

% end

fitBestR(j) = fitR(j);

elseif size(I,3) == 3

%RGB image

fitR(j) = sum(probR(1:xR(j,1))) * (sum((1:xR(j,1)) .* probR(1:xR(j,1)) / sum(probR(1:xR(j,1)))) - sum((1:Lmax) .* probR(1:Lmax))) ^ 2;

for jlevel = 2:level - 1

end

fitBestR(j) = fitR(j);

fitG(j) = sum(probG(1:xG(j,1))) * (sum((1:xG(j,1)) .* probG(1:xG(j,1)) / sum(probG(1:xG(j,1)))) - sum((1:Lmax) .* probG(1:Lmax))) ^ 2;

for jlevel = 2:level - 1

fitG(j) = fitG(j) + sum(probG(xG(j,jlevel - 1) + 1:xG(j,jlevel))) * (sum((xG(j,jlevel - 1) + 1:xG(j,jlevel)) .* probG(xG(j,jlevel - 1) + 1:xG(j,jlevel)) / sum(probG(xG(j,jlevel - 1) + 1:xG(j,jlevel)))) - sum((1:Lmax) .* probG(1:Lmax))) ^ 2;

end

fitG(j) = fitG(j) + sum(probG(xG(j,level - 1) + 1:Lmax)) * (sum((xG(j,level-1) + 1:Lmax) .* probG(xG(j,level - 1) + 1:Lmax) / sum(probG(xG(j,level - 1) + 1:Lmax))) - sum((1:Lmax) .* probG(1:Lmax))) ^ 2;

fitBestG(j) = fitG(j);

fitB(j) = sum(probB(1:xB(j,1))) * (sum((1:xB(j,1)) .* probB(1:xB(j,1)) / sum(probB(1:xB(j,1)))) - sum((1:Lmax) .* probB(1:Lmax))) ^ 2;

for jlevel = 2:level - 1

fitB(j) = fitB(j) + sum(probB(xB(j,jlevel - 1) + 1:xB(j,jlevel))) * (sum((xB(j,jlevel - 1) + 1:xB(j,jlevel)) .* probB(xB(j,jlevel - 1) + 1:xB(j,jlevel)) / sum(probB(xB(j,jlevel - 1) + 1:xB(j,jlevel)))) - sum((1:Lmax) .* probB(1:Lmax))) ^ 2;

end

fitB(j) = fitB(j) + sum(probB(xB(j,level - 1) + 1:Lmax)) * (sum((xB(j,level - 1) + 1:Lmax) .* probB(xB(j,level - 1) + 1:Lmax) / sum(probB(xB(j,level - 1) + 1:Lmax))) - sum((1:Lmax) .* probB(1:Lmax))) ^ 2;

fitBestB(j) = fitB(j);

end

if size(I,3) == 1

%Imagen escala de Grises

fitR = fitR';

fitBestR = fitBestR';

elseif size(I,3) == 3

% Imagen RGB

fitR = fitR';

fitBestR = fitBestR';

fitG = fitG';

fitBestG = fitBestG';