利用MATLAB代码实现图像的分类K-means算法

在MATLAB中实现K-means图像分类算法，通常会分为以下几个步骤：

1. **加载和预处理图像**:

% 加载图像数据
img = imread('your_image.jpg');
% 将RGB图像转换为灰度图像，如果需要
gray_img = rgb2gray(img);
% 对图像进行归一化
normalized_img = im2double(gray_img) / 255;

2. **将图像转为向量表示**:

% 将像素值展平成行向量
pixel_vectors = reshape(normalized_img, [], 1);

3. **选择初始质心（中心点）**:

% 初始化k个随机质心
k = 5; % 根据类别数设置k值
centroids = rand(k, size(pixel_vectors, 2));

4. **迭代K-means过程**:

% 设置最大迭代次数和误差阈值
max_iters = 100;
tolerance = 1e-6;

for iter = 1:max_iters
    % 分配每个像素到最近的质心
    assignments = pdist2(pixel_vectors, centroids, 'euclidean');
    clusters = kmeans(pixel_vectors, k, 'InitialCenters', centroids, 'EmptyAction', 'singleton');

    % 更新质心
    new_centroids = zeros(size(centroids));
    for i = 1:k
        new_centroids(i,:) = mean(pixel_vectors(clusters == i,:), 1);
    end
    
    % 检查停止条件：质心变化小于指定阈值
    if norm(new_centroids - centroids, 'fro') < tolerance
        break;
    end
    centroids = new_centroids;
end

5. **结果可视化**:

% 绘制聚类后的结果
figure;
imagesc(normalized_img);
hold on;
colors = jet(k);
for i = 1:k
    cluster_pixels = pixel_vectors(find(clusters == i), :);
    plot(cluster_pixels(:,2), cluster_pixels(:,1), '.', 'Color', colors(i,:));
end
hold off;
title(['K-means Clustering Result (k = ' num2str(k) ')']);