matlab实现一次性实现多个文件夹图片转化为.mat文件

%这里是主函数：命名为readImg.m;

clc;clear;
%---read_image;
filepath = 'G:\人脸重建\data\src_all\';%图片路径可以根据自己需要修改;
pic = '*jpg';%图片格式，根据自己的图片格式进行修改;
database = build_database_src(filepath,pic);

%build_database_src.m子函数;

function database = build_database_src(rt_data_dir,suffix)

fprintf('dir the database');
subfolders = dir(rt_data_dir);
 
database = [];
 
database.imnum = 0;
database.cname = {};
database.label = [];
database.path = [];
database.nclass = 0;
label = 0;
 
k = 1;
src_x = [];
src_y = [];
 
for ii = 1 : length(subfolders)
    subname = subfolders(ii).name;
    
    if ~strcmp(subname,'.') & ~strcmp(subname,'..')
        
        database.nclass = database.nclass + 1;
        database.cname{database.nclass} = subname;
        
        frames = dir(fullfile(rt_data_dir,subname,suffix));
        c_num = length(frames);
        
        database.imnum = database.imnum + c_num;
        database.label = [database.label;ones(c_num,1) * database.nclass];
        
        label = numel(database.cname);
        
        for jj = 1 : c_num
            fprintf('folder : %d , num : %d \n',ii - 1,jj);
            
            kk1 = frames(jj).name;%显示具体图片名;
            c_path = fullfile(rt_data_dir,subname,frames(jj).name);
            database.path = [database.path,c_path];
            img = imread(database.path);%读取图片;
            im = imresize(img,[64,64]);%设置图片大小;
            
            if(length(size(im)) == 3)%如果是彩色图进行灰度化(根据个人需要);
                gray = rgb2gray(im);
            else
                gray = im;
            end
            
            gray = double(gray) / 255;%对图片数据做归一化(根据个人需要);
            database.path = [];
            
            src_x(:,:,k) = gray;%存储图像数据;
            src_y(1,k) = label;%给数据加标签;
            k = k + 1;
        end
    end
    
end
 
%保存转化好的.mat文件;
file = ['src_x.mat'];
save(file,'src_x','-mat');
file = ['src_y.mat'];
save(file,'src_y','-mat');
 
end

我的路径：G:\人脸重建\data\src_all。运行程序可以看到生成两个.mat文件。

参考：https://blog.csdn.net/yongjiankuang/article/details/72824081