机器学习的相关算法的实现需要大量使用到矩阵运算。用C++实现的话,OpenCv里面的Mat数据结构是一个很好的选择。
初始化
- 创建cv::Mat,并且初始化cv::Scalar::all(0)
- cv::Mat 可以直接使用cout进行输出
cv::Mat a(3, 3, CV_32FC1, cv::Scalar::all(0));
cout << a << endl;
使用数组初始化
int bArray[3][3] = { {1, 2, 3}, {4, 5, 6}, {7, 8, 9} };
cv::Mat b(3, 3, CV_8UC1, bArray);
PS: CV_8UC1代表的含义(摘抄自这里)
CV_<bit_depth>(S|U|F)C<number_of_channels>
1--bit_depth---比特数---代表8bite,16bites,32bites,64bites---举个例子吧--比如说,如
如果你现在创建了一个存储--灰度图片的Mat对象,这个图像的大小为宽100,高100,那么,现在这张
灰度图片中有10000个像素点,它每一个像素点在内存空间所占的空间大小是8bite,8位--所以它对
应的就是CV_8
2--S|U|F--S--代表---signed int---有符号整形
U--代表--unsigned int--无符号整形
F--代表--float---------单精度浮点型
3--C<number_of_channels>----代表---一张图片的通道数,比如:
1--灰度图片--grayImg---是--单通道图像
2--RGB彩色图像---------是--3通道图像
3--带Alph通道的RGB图像--是--4通道图像
.create()函数
可以创建规定大小的Mat
然后使用.at<_type>(i, j)赋值
cv::Mat c;
c.create(3, 3, CV_32FC1);
c.at<float>(0, 0) = 1;
c.at<float>(0, 1) = 2;
c.at<float>(0, 2) = 3;
c.at<float>(1, 0) = 4;
c.at<float>(1, 1) = 5;
c.at<float>(1, 2) = 6;
c.at<float>(2, 0) = 7;
c.at<float>(2, 1) = 8;
c.at<float>(2, 2) = 9;
cout << "c" << endl << c << endl;
下面是.at<type>的对应表
Mat---------CV_8U
Mat-----------CV_8S
Nat_---------CV_16S
Mat_--------CV_16U
Mat_-----------CV_32S
Mat_----------CV_32F
Mat_--------CV_64F
cv::Scalar = cv::sum(cv::Mat scr)
求和函数
src中所有的值求和
cv::Scalar sc = cv::sum(c);
cout << sc[0] << endl; //输出和!!!
如果按照某个维度来求和,就需要使用**reduce()**函数
//1 表示按行求和
//0 表示按列求和
cv::reduce(c, cc, 1, CV_REDUCE_SUM);
cout << "cc" << endl << cc << endl;
cout << c.dims << endl; //输出维度
cout << c.rows << endl; //输出行数
cout << c.cols << endl; //输出列数
深浅复制问题
直接用 = 赋值的话,就是传引用而已
.clone() 是拷贝一份新的
cv::Mat copyC = c;
copyC.at<float>(0, 0) = 10;
cout << "c:" << endl << c << endl; //c也被更改了
cout << "copyC:" << endl << copyC << endl;
c.at<float>(0, 0) = 1;
cv::Mat copyC1 = c.clone();
copyC1.at<float>(0, 0) = 10;
cout << "c:" << endl << c << endl; //c未被更改了
cout << "copyC1:" << endl << copyC1 << endl;
.copyTo(dst)
for (int i = 0; i < c.cols; i++)
{
cc.copyTo(c.col(i)); //把cc(列向量)复制到c的第i列上面
}
cout << c << endl;
cc = cc.t(); //矩阵转置
for (int i = 0; i < c.rows; i++)
{
cc.copyTo(c.row(i)); //cc是列向量,不可以为行向量赋值。但是加上cc = cc.t(); 之后就可以了
}
cout << c << endl;
所有的调试代码;
以后学习到了在加
#include <opencv2/opencv.hpp>
#include <iostream>
using namespace std;
using namespace cv;
int main()
{
cv::Mat a(3, 3, CV_32FC1, cv::Scalar::all(0));
cout << "a" << endl;
cout << a << endl;
int bArray[3][3] = { {1, 2, 3}, {4, 5, 6}, {7, 8, 9} };
cv::Mat b(3, 3, CV_8UC1, bArray);
cout << "b" << endl;
cout << b << endl;
cv::Mat c;
c.create(3, 3, CV_32FC1);
c.at<float>(0, 0) = 1;
c.at<float>(0, 1) = 2;
c.at<float>(0, 2) = 3;
c.at<float>(1, 0) = 4;
c.at<float>(1, 1) = 5;
c.at<float>(1, 2) = 6;
c.at<float>(2, 0) = 7;
c.at<float>(2, 1) = 8;
c.at<float>(2, 2) = 9;
cout << "c" << endl << c << endl;
//cv::Mat cc;
cv::Scalar sc = cv::sum(c);
cout << "sum C = " << sc[0] << endl;
//reduce()
cv::Mat cc;
//1 表示按行求和
//0 表示按列求和
cv::reduce(c, cc, 1, CV_REDUCE_SUM);
cout << "cc" << endl << cc << endl;
cout << c.dims << endl; //输出维度
cout << c.rows << endl; //输出行数
cout << c.cols << endl; //输出列数
//复制
cv::Mat copyC = c;
copyC.at<float>(0, 0) = 10;
cout << "c:" << endl << c << endl; //c也被更改了
cout << "copyC:" << endl << copyC << endl;
c.at<float>(0, 0) = 1;
cv::Mat copyC1 = c.clone();
copyC1.at<float>(0, 0) = 10;
cout << "c:" << endl << c << endl; //c未被更改了
cout << "copyC1:" << endl << copyC1 << endl;
for (int i = 0; i < c.cols; i++)
{
cc.copyTo(c.col(i)); //把cc(列向量)复制到c的第i列上面
}
cout << c << endl;
cc = cc.t();
for (int i = 0; i < c.rows; i++)
{
cc.copyTo(c.row(i)); //cc是列向量,不可以为行向量赋值
}
cout << c << endl;
//int tt;
//cin >> tt;
}