【一】原理
1、开启两个摄像头，系统默认从0开始查找，两个摄像头使用0和1就够啦，然后把显示窗口尺寸固定，这里用的是320和240

VideoCapture capture1(0);
VideoCapture capture2(1);

2、创建一个黑色图片，大小刚好可以容纳两个摄像头窗口，width=640，height=240

Mat frame(240,640,CV_8UC3,Scalar(0,0,0));

3、创建ROI区域

ROIImage1 = frame(Rect(0,0,320,240));
ROIImage2 = frame(Rect(320, 0, 320, 240));

4、创建掩膜

掩膜必须是灰度图，因此先要把图片转化为灰度图

cvtColor(frame1, gray_Image1,CV_RGB2GRAY);
cvtColor(frame2, gray_Image2, CV_RGB2GRAY);

然后把图片复制给掩膜

mask1 = gray_Image1;
mask2 = gray_Image2;

5、把掩膜复制到ROI区域，frame1为要覆盖的图片，意思就是把frame1复制ROI区域的掩膜上。

frame1.copyTo(ROIImage1, mask1);
frame2.copyTo(ROIImage2, mask2);

【二】效果图

【三】源码

#include <iostream>
#include <opencv2/opencv.hpp>

using namespace std;
using namespace cv;

//【画框】
void Draw_rectangle(Mat img,Point pt1, Point pt2)
{
    rectangle(img, pt1, pt2, Scalar(0, 255, 0), 2, 8, 0);
}

int main(int argc, char *argv[])
{

    Mat srcImage,srcImage2,srcImage3,srcImage4;
    Mat ROIImage1,ROIImage2,ROIImage3,ROIImage4,mask1,mask2,mask3,mask4,gray_Image1, gray_Image2,gray_Image3,gray_Image4;
    VideoCapture cap(0);
    VideoCapture cap1("car_input.mp4");
    VideoCapture cap2("carMove2.mp4");
    VideoCapture cap3("carMove.mp4");
    cap.set(CV_CAP_PROP_FRAME_WIDTH, 320);
    cap.set(CV_CAP_PROP_FRAME_HEIGHT, 240);
    cap1.set(CV_CAP_PROP_FRAME_WIDTH, 320);
    cap1.set(CV_CAP_PROP_FRAME_HEIGHT, 240);
    cap2.set(CV_CAP_PROP_FRAME_WIDTH, 320);
    cap2.set(CV_CAP_PROP_FRAME_HEIGHT, 240);
    cap3.set(CV_CAP_PROP_FRAME_WIDTH, 320);
    cap3.set(CV_CAP_PROP_FRAME_HEIGHT, 240);

    if (!cap.isOpened())
    {
        cout << "VideoCapture open fail" << endl;
        return 0;
    }
    if (!cap1.isOpened())
    {
        cout << "VideoCapture1 open fail" << endl;
        return 0;
    }

    if (!cap2.isOpened())
    {
        cout << "VideoCapture2 open fail" << endl;
        return 0;
    }
    if (!cap3.isOpened())
    {
        cout << "VideoCapture3 open fail" << endl;
        return 0;
    }
    Mat frame3(480,640,CV_8UC3,Scalar(0,0,0));
    char temp_1[20];
    char temp_2[20];
    char temp_3[20];
    char temp_4[20];
    while (cap.isOpened())
    {
        cap >> srcImage;
        cap1 >> srcImage2;
        cap2 >> srcImage3;
        cap3 >> srcImage4;

        resize(srcImage2, srcImage2,Size(320,240));
        resize(srcImage3, srcImage3,Size(320,240));
        resize(srcImage4, srcImage4,Size(320,240));
        Draw_rectangle(srcImage, Point(0,0),Point(srcImage.cols, srcImage.rows));
        Draw_rectangle(srcImage2, Point(0, 0), Point(srcImage2.cols, srcImage2.rows));
        Draw_rectangle(srcImage3, Point(0, 0), Point(srcImage3.cols, srcImage3.rows));
        Draw_rectangle(srcImage4, Point(0, 0), Point(srcImage4.cols, srcImage4.rows));

        sprintf(temp_1, "capture1");
        putText(srcImage, temp_1, Point(srcImage.cols / 16, srcImage.rows / 10), FONT_HERSHEY_SIMPLEX, 0.8, Scalar(255, 0, 255));
        sprintf(temp_2, "capture2");
        putText(srcImage2, temp_2, Point(srcImage2.cols / 16, srcImage2.rows / 10), FONT_HERSHEY_SIMPLEX, 0.8, Scalar(255, 0, 255));
        sprintf(temp_3, "capture3");
        putText(srcImage3, temp_3, Point(srcImage3.cols / 16, srcImage3.rows / 10), FONT_HERSHEY_SIMPLEX, 0.8, Scalar(255, 0, 255));
        sprintf(temp_4, "capture4");
        putText(srcImage4, temp_4, Point(srcImage4.cols / 16, srcImage4.rows / 10), FONT_HERSHEY_SIMPLEX, 0.8, Scalar(255, 0, 255));


        ROIImage1 = frame3(Rect(0,0,320,240));
        ROIImage2 = frame3(Rect(320, 0, 320, 240));
        ROIImage3 = frame3(Rect(0, 240, 320, 240));
        ROIImage4 = frame3(Rect(320, 240, 320, 240));
        //imshow("ROIImage1", ROIImage1);
        //imshow("ROIImage2", ROIImage2);
        cvtColor(srcImage, gray_Image1,CV_RGB2GRAY);
        cvtColor(srcImage2, gray_Image2, CV_RGB2GRAY);
        cvtColor(srcImage3, gray_Image3, CV_RGB2GRAY);
        cvtColor(srcImage4, gray_Image4, CV_RGB2GRAY);

        mask1 = gray_Image1;
        mask2 = gray_Image2;
        mask3 = gray_Image3;
        mask4 = gray_Image4;
        srcImage.copyTo(ROIImage1, mask1);
        srcImage2.copyTo(ROIImage2, mask2);
        srcImage3.copyTo(ROIImage3, mask3);
        srcImage4.copyTo(ROIImage4, mask4);
//        imshow("ROIImage1", srcImage);
//        imshow("ROIImage2", srcImage2);
//        imshow("srcImage3", srcImage3);
//        imshow("srcImage4", srcImage4);
        imshow("main", frame3);
        if (waitKey(25)==27)
        {
            break;
        }
    }
    cap.release();
    destroyAllWindows();
    return 0;
}

通过hstack, vstack拼接

代码

import cv2
import numpy as np
cap = cv2.VideoCapture(0)
cap_t = cv2.VideoCapture(1)
while cap.isOpened():
	ret, frame = cap.read()
	ret_t, frame_t = cap_t.read()
	frame_h = np.hstack((frame, frame_t)) # frame_v = np.vstack((frame, frame_t))
	cv2.imshow('title', frame_h)
	if cv2.waitKey(10) == 113: #点击q的时候退出
		break

效果（hstack）：

hstack是横向拼接，vstack是纵向拼接

通过一个大的矩阵显示

代码

import cv2
import numpy as np

frame_h = np.zeros((480, 1280, 3), dtype=np.uint8) # 创建一个1280*640的矩阵用来显示
cap = cv2.VideoCapture(0)
cap_t = cv2.VideoCapture(1)
while cap.isOpened():
	ret, frame = cap.read()
	ret_t, frame_t = cap_t.read()
	frame_h[0:480, 0:640, :] = frame
	frame_h[0:480, 640:1280, :] = frame_t
	cv2.imshow('title', frame_h)
	if cv2.waitKey(10) == 113:
		break

效果

注意，再创建大的矩阵的使用要填入dtype=np.uint8不然会出现显示白屏或是显示色彩不对的情况

总结

第一种方法适用于摄像头或是图片数量固定的时候，通过hstack和vstack拼接每次仅能两张图片拼接。第二种方法适用于摄像头数量不定的情况，设置好大的显示矩阵后，只需将你想要显示的区域填入内容就可以显示，其他部分会显示黑色。

经常性的图像处理用到一个摄像头就行了，但是有时候也需要用到多个摄像头，这里介绍下如何打开多个摄像头。这里介绍三种打开方式 opencv ,MFC,QT,
其实QT和MFC都是引用opencv的库总体上程序没有太多的变化。。
介绍下QT打开三个摄像头，，，，，，
VideoCapture cap1;
VideoCapture cap2;
VideoCapture cap3;
Mat frameleft,frameright,framecenture,framelefthsv,framegray;
cap1.open(0);
cap2.open(1);
cap3.open(2);
while(1)
{
cap1.read(frameleft);
cap2.read(frameright);
cap3.read(framecenture);
cvtColor(frameleft,framelefthsv,COLOR_BGR2HSV);
cvtColor(frameright,framegray,COLOR_BGR2GRAY);

    QImage imgleft=mat2QImage(framelefthsv);
    QImage imgright=mat2QImage(framegray);
    QImage imgcen=mat2QImage(framecenture);

    ui->label->setScaledContents(true);
    ui->label_2->setScaledContents(true);
    ui->label_3->setScaledContents(true);

     ui->label->setPixmap(QPixmap::fromImage(imgleft));
     ui->label_2->setPixmap(QPixmap::fromImage(imgright));
     ui->label_3->setPixmap(QPixmap::fromImage(imgcen));
     waitKey(30);

MFC…
opencv…
好吧，原本想再写个EMGUCV打开多个摄像头的，，突然发现4.0的和3.0的不太一样了，，，以前用的是3.0的。。。要再看看了