From[ http://blog.csdn.net/ab1322583838/article/details/52054260]
This blog has a clear explanation on color processing. It is also the place I should learn when I write a blog in the future.

I. Goals

Learning the use of Mat and the method of assigning initial values to Mat images

2. Function Description

Mat red_img(Size(), CV_8UC3, Scalar());
First parameter: Set the size of the image matrix
Second parameter: Set the type of image, CV_8UC3 refers to an 8-bit unsigned integer three-channel matrix
The third parameter: represents an array with four elements. This type is widely used in OpenCV to transfer pixel values. In this section, we will further use it to represent RGB color values (three parameters). If you do not use the fourth parameter, you do not need to define it.
Let's take an example, if we give the following color parameter expressions:
Scalar( a, b, c )
Then the RGB color values defined are: Red = c, Green = b, Blue= a

3. Program Code

#include "cv.h"                             //  OpenCV File Header
#include "highgui.h"
#include "cvaux.h"
#include "cxcore.h"
#include "opencv2/opencv.hpp"
#include "opencv2/imgproc.hpp"
#include <iostream>
#include <string>

using namespace cv;
using namespace std;

int
main(int argc, char *argv[])
{
    // Initialization of image matrix
    Mat red_img(Size(480, 480), CV_8UC3, Scalar(0, 0, 255));//Red image
    Mat white_img(Size(480, 480), CV_8UC3, Scalar::all(255));//White image
    Mat black_img = Mat::zeros(Size(480, 480), CV_8UC3);//Black image

    // Re-assigning the channel of the image
    Mat green_img = red_img.clone();//The purpose of this operation is to obtain pixels of the same size as red pictures.
    green_img = Scalar(0, 255, 0);//Green image

    namedWindow("red image", CV_WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);// Create a window
    namedWindow("white image", CV_WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);// Create a window
    namedWindow("black image", CV_WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);// Create a window
    namedWindow("green image", CV_WINDOW_AUTOSIZE | CV_WINDOW_FREERATIO);// Create a window
    imshow("red image", red_img);// Display pictures in Windows
    imshow("white image", white_img);// Display pictures in Windows
    imshow("black image", black_img);// Display pictures in Windows
    imshow("green image", green_img);// Display pictures in Windows
    waitKey(0);// Waiting for a button to close the program
}

IV. Interpretation

1. Red image

Mat red_img(Size(480, 480), CV_8UC3, Scalar(0, 0, 255));

2. White image

Mat white_img(Size(480, 480), CV_8UC3, Scalar::all(255));

Size(480, 480) is an image matrix that defines a 480 x 480. CV_8UC3 is an 8-bit unsigned integer three-channel matrix. Scalar::all(255) means that in these three channels, channel B is 255, channel G is 255, and channel R is 255.
3. Black image

Mat black_img = Mat::zeros(Size(480, 480), CV_8UC3);

zeros() denotes the initialization of a black image matrix with all values of 0.
It is equivalent to

Mat black_img(Size(480, 480), CV_8UC3, Scalar::all(0))；
Mat black_img(Size(480, 480), CV_8UC3, Scalar(0,0,0))；

4. Green Image

Mat green_img = red_img.clone();
green_img = Scalar(0, 255, 0);//Green image

Mat green_img = red_img.clone(); denotes that green_img is the same as red_img and is cloned by red_img.
green_img = Scalar(0, 255, 0); means to reassign the three channels, in which B channel is 0, G channel is 255, R channel is 0.
V. RESULTS

The results are as follows: