Opencv C++ three ways to access pixels in the image

Procedure description

// Program description: Three methods for OpenCV C++ to access pixels in images
// Reference: Mao Xingyun "OpenCV3 Programming Introduction"
// Operating system: Windows 10 64bit
// Development language: C++
// IDE version: Visual Studio 2019
// OpenCV version: 4.20

Any image processing algorithm starts with operating each pixel.
OpenCV provides three algorithms for accessing each pixel:
1. Pointer access
2. Iterator
3. Dynamic address calculation

The difference between these three methods is very obvious in debug mode, but in release mode, the difference is not obvious. In order to improve access efficiency, we use the colorReduce function; to reduce the color space to reduce the color work.

Method one uses pointers to access pixels

#include <opencv2/core/core.hpp>  
#include <opencv2/highgui/highgui.hpp>  
#include <iostream>  
using namespace std;
using namespace cv;

//-----------------------------------【全局函数声明部分】-----------------------------------
//          描述：全局函数声明
//-----------------------------------------------------------------------------------------------
void colorReduce(Mat& inputImage, Mat& outputImage, int div);



//--------------------------------------【main( )函数】---------------------------------------
//          描述：控制台应用程序的入口函数，我们的程序从这里开始执行
//-----------------------------------------------------------------------------------------------
int main()
{
	//【1】创建原始图并显示
	Mat srcImage = imread("100.jpg");
	imshow("原始图像", srcImage);

	//【2】按原始图的参数规格来创建创建效果图
	Mat dstImage;
	dstImage.create(srcImage.rows, srcImage.cols, srcImage.type());//效果图的大小、类型与原图片相同 

	//【3】记录起始时间
	double time0 = static_cast<double>(getTickCount());  //记录开始时间

	//【4】调用颜色空间缩减函数
	colorReduce(srcImage, dstImage, 32);

	//【5】计算运行时间并输出
	time0 = ((double)getTickCount() - time0) / getTickFrequency();   //结束时间-开始时间，并化为秒单位
	cout << "\t此方法运行时间为： " << time0 << "秒" << endl;  //输出运行时间

	//【6】显示效果图
	imshow("效果图", dstImage);
	waitKey(0);
}


//---------------------------------【colorReduce( )函数】---------------------------------
//          描述：使用【指针访问：C操作符[ ]】方法版的颜色空间缩减函数
//----------------------------------------------------------------------------------------------
void colorReduce(Mat& inputImage, Mat& outputImage, int div)
{
	//参数准备
	outputImage = inputImage.clone();  //拷贝实参到临时变量
	int rowNumber = outputImage.rows;  //行数
	int colNumber = outputImage.cols * outputImage.channels();  //列数 x 通道数=每一行元素的个数

	//双重循环，遍历所有的像素值
	for (int i = 0; i < rowNumber; i++)  //行循环
	{
		uchar* data = outputImage.ptr<uchar>(i);  //获取第i行的首地址
		for (int j = 0; j < colNumber; j++)   //列循环
		{
			// ---------【开始处理每个像素】-------------     
			data[j] = data[j] / div * div + div / 2;
			// ----------【处理结束】---------------------
		}  //行处理结束
	}
}

running result

Method 2: Use iterators to access pixels

//-------------------------------------【colorReduce( )函数】-----------------------------
//		描述：使用【迭代器】方法版的颜色空间缩减函数
//----------------------------------------------------------------------------------------------
void colorReduce(Mat& inputImage, Mat& outputImage, int div)  
{  
	//参数准备
	outputImage = inputImage.clone();  //拷贝实参到临时变量
	//获取迭代器
	Mat_<Vec3b>::iterator it = outputImage.begin<Vec3b>();  //初始位置的迭代器
	Mat_<Vec3b>::iterator itend = outputImage.end<Vec3b>();  //终止位置的迭代器

	//存取彩色图像像素
	for(;it != itend;++it)  
	{  
		// ------------------------【开始处理每个像素】--------------------
		(*it)[0] = (*it)[0]/div*div + div/2;  
		(*it)[1] = (*it)[1]/div*div + div/2;  
		(*it)[2] = (*it)[2]/div*div + div/2;  
		// ------------------------【处理结束】----------------------------
	}  
}  

running result

Method two: use dynamic address calculation with at to access pixels

//----------------------------------【colorReduce( )函数】-------------------------------
//          描述：使用【动态地址运算配合at】方法版本的颜色空间缩减函数
//----------------------------------------------------------------------------------------------
void colorReduce(Mat& inputImage, Mat& outputImage, int div)  
{  
	//参数准备
	outputImage = inputImage.clone();  //拷贝实参到临时变量
	int rowNumber = outputImage.rows;  //行数
	int colNumber = outputImage.cols;  //列数

	//存取彩色图像像素
	for(int i = 0;i < rowNumber;i++)  
	{  
		for(int j = 0;j < colNumber;j++)  
		{  	
			// ------------------------【开始处理每个像素】--------------------
			outputImage.at<Vec3b>(i,j)[0] =  outputImage.at<Vec3b>(i,j)[0]/div*div + div/2;  //蓝色通道
			outputImage.at<Vec3b>(i,j)[1] =  outputImage.at<Vec3b>(i,j)[1]/div*div + div/2;  //绿色通道
			outputImage.at<Vec3b>(i,j)[2] =  outputImage.at<Vec3b>(i,j)[2]/div*div + div/2;  //红是通道
			// -------------------------【处理结束】----------------------------
		}  // 行处理结束     
	}  
}  

running result

to sum up

Comparing the three methods, the first one is the fastest.