用MATLAB针对Feret人脸库,编写一个基于主成分分析的人脸识别函数,输入一张人脸图像,识别出库中的人脸,并用左右子图显示结果。要求1.函数的主体算法为主成分分析算法2.每个人的7张人脸图像中,选择6张图像作为训练样本,1

张作为测试样本，计算测试样本与每个人训练样本的距离，选出距离最小的作为识别结果。以下是MATLAB代码实现：

function [recognized_face, eigenfaces, weights] = face_recognition(training_folder, test_image)

% 读取训练数据
training_files = dir(fullfile(training_folder, '*.jpg'));
num_train = length(training_files);
train_images = zeros(3600, num_train);
for i = 1:num_train
    img = imread(fullfile(training_folder, training_files(i).name));
    train_images(:, i) = img(:);
end

% 计算平均脸
mean_face = mean(train_images, 2);

% 计算特征向量
A = train_images - mean_face;
S = A' * A / (num_train - 1);
[U, D, ~] = svd(S);
eigenfaces = A * U;

% 计算权重
test_image = double(test_image(:)) - mean_face;
weights = eigenfaces' * test_image;

% 识别人脸
min_distance = inf;
recognized_face = '';
for i = 1:num_train
    dist = norm(weights - eigenfaces(:, i)' * A(:, i));
    if dist < min_distance
        min_distance = dist;
        recognized_face = training_files(i).name;
    end
end

% 显示结果
figure;
subplot(1, 2, 1);
imshow(reshape(test_image + mean_face, 60, 60));
title('测试图像');
subplot(1, 2, 2);
imshow(fullfile(training_folder, recognized_face));
title('识别结果');

end

其中，`training_folder`为训练数据所在文件夹的路径，`test_image`为待识别的测试图像。函数返回识别结果`recognized_face`、特征向量矩阵`eigenfaces`和权重向量`weights`。函数主体算法为主成分分析算法，使用奇异值分解（SVD）计算特征向量。每个人的7张人脸图像中，选择6张图像作为训练样本，1张作为测试样本，计算测试样本与每个人训练样本的距离，选出距离最小的作为识别结果。最后用左右子图显示结果。