|
%%%%%%%%%%%%%%% FCM算法分割图像 %%%%%%%%%%%%%%
function clusterResult = FCM(imagePath, C, V, M, iter, epsm)
% 模糊C均值(FCM)聚类算法分割图像
% clusterResult = FCM(imagePath, C, V, M, iter, epsm)
% Example: clusterResult = FCM('E:\Image\lena.bmp')
% clusterResult = FCM('E:\Image\lena.bmp',3,[0 127 255])
% Input:
% imagePath: 图像路径
% C: 类别数，缺省值为2
% V: 初始化聚类中心,缺省值为[0 255]
% M: 加权指数,缺省值为2
% iter: 迭代次数,缺省值为100
% epsm: 迭代停止阈值,缺省值为1.0e-2
% Output:
% clusterResult: 聚类中心结果
% Note:
% C的值要与V的初始化聚类中心个数相同
% 设定缺省值
if nargin < 6
epsm = 1.0e-2;
end
if nargin < 5
iter = 100;
end
if nargin < 4
M = 2;
end
if nargin < 3
V = [0 255];
end
if nargin < 2
C = 2;
end
% 读入图像及其信息
I = imread(imagePath);
figure, imshow(I);
title('原图像');
[row col] = size(I);
grayHist = imhist(I);
figure, imhist(I);
title('直方图');
histProb = grayHist / (row * col);
len = length(histProb);
tic
% FCM迭代过程
cnt = 0;
while(cnt < iter)
% 计算隶属度函数(注意要特殊考虑某个像素点和聚类中心一样的情况)
for i = 1 : len
flag = 0;
for j = 1 : C
if i == V(j)
U(j, i) = 1.0;
if j == 1
U(j + 1 : C, i) = 0.0;
elseif j == C
U(1 : C - 1, i) = 0.0;
else
| 
转播
