I have 25 images in a folder and I want to make a .mat file for a road-sign recognition system.
What are the steps for making a .mat file in Matlab?
解决方案
There may be a better approach for images but here is what I know. If you want to control what goes into the .mat file you can specify what variables in your workspace will be saved using the save command.
% MATLAB R2017a
X = rand(273,273);
Y = rand(273,273);
Z = rand(273,273);
save FileName X Y Z
This creates a file FileName.mat.
You can access the contents using the load command.
clear
load FileName
To save everything in the workspace to a .mat file, use the save command without specifying the variables to save (MATLAB will then save them all).
W = rand(273,273);
save FileName
See the linked documentation for more options and examples.
This requires you to loop through the images in the folder. A direct approach to this is directly loading the images using a loop over [filepath 'image' num2str(j) '.jpg'] with index j where filepath = 'C:\Users\user1\Folder\ImageFolder\'. This uses string concatenation and the num2str command.
If you need to change your current directory within the script,the cd function is useful.

#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include <iostream>
int main( )
{
  cv::Mat image1; 
 
  cv::Mat image2( 6, 6, CV_8UC1 );
  
  cv::Mat image3( cv::Size(7,7), CV_8UC3 );
  
  cv::Mat image4( 8, 8, CV_32FC2, cv::Scalar(1,3) );
  
  cv::Mat image5( cv::Size(9,9), CV_8UC3, cv::Scalar(1,2,3) );
  
  cv::Mat image6( image2 );
  
  std::cout << image1 << std::endl;
  getchar();
  std::cout << image2 << std::endl;
  getchar();
  std::cout << image3 << std::endl;
  getchar();
  std::cout << image4 << std::endl;
  getchar();
  std::cout << image5 << std::endl;
  getchar();
  std::cout << image6 << std::endl;
  return 0;
}

只用来记录学习笔记
新生成一个rgb为（127, 0, 255）的图像：
	Mat dst;
	dst=Mat(src.size(), src.type()); //和src图片一样大小，一样类型
	dst = Scalar(127, 0, 255);
	namedWindow("output", CV_WINDOW_AUTOSIZE);
	imshow("output", dst);

克隆图片方法1：
Mat dst = src.clone();
namedWindow("output", CV_WINDOW_AUTOSIZE);
imshow("output", dst);

克隆图片方法2：
Mat dst;
src.copyTo(dst);
namedWindow("output", CV_WINDOW_AUTOSIZE);
imshow("output", dst);

颜色空间转换函数：
cvtColor(src, dst, CV_BGR2GRAY); //转成灰色

转成灰色效果图：


通道：
src.channels()


基本上，描述一个像素点，如果是灰度，那么只需要一个数值来描述它，就是单通道。

如果一个像素点，有RGB三种颜色来描述它，就是三通道。

windows的bmp有时候是一个四通道图像，R、G、B加上一个A通道，表示透明度。

获取图像数据指针：
const uchar* firstRow = dst.ptr<uchar>(0);

新建一个Mat：
Mat  m1;
m1.create(src.size(), src.type());
m1 = Scalar(0, 0, 255);

新建一个Mat：
Mat m2 = Mat::zeros(2,2 ,CV_8UC1);

新建Mat：
Mat C = (Mat_<double>(3, 3) << 0, -1, 0, -1, 5, -1, 0, -1, 0);

//整理创建Mat的方法
	//1. 使用Mat()构造函数构建
	/** @overload
	@param rows Number of rows in a 2D array.    二维数组的行数，又图像的高度
	@param cols Number of columns in a 2D array. 二维数组的列数，又图像的宽度
	@param type Array type. Use CV_8UC1, ..., CV_64FC4 to create 1-4 channel matrices, or
	CV_8UC(n), ..., CV_64FC(n) to create multi-channel (up to CV_CN_MAX channels) matrices. 数组的类型，又图片的类型
	*/
	//Mat(int rows, int cols, int type);
	Mat m(120, 120, CV_8UC3); //创建120×120的三通道图像，图像的像素值随机生成
	randu(m, Scalar::all(0), Scalar::all(255));//也可以自己指定随机值的期望范围
	Mat m2(120, 120, CV_8UC3, Scalar(0, 255, 0));//创建120×120的三通道图像，指定图像的像素值

	//2. 采用Matlab式初始化
	Mat eyeImg = Mat::eye(4, 4, CV_8UC3);//创建对角线为1，其余为0的图像
	Mat oneImg = Mat::ones(4, 4, CV_8UC3);//创建像素值全为1的图像
	Mat zeroImg = Mat::zeros(4, 4, CV_8UC3);//创建像素值全为0的图像

	//3. 从已有的图像处拷贝
	Mat cloneImg = eyeImg.clone();
	Mat copyImg;
	eyeImg.copyTo(copyImg);