双目测距步骤四：深度图+内外参数+画布

最新推荐文章于 2024-06-20 14:37:31 发布

Linux丶狂热者

最新推荐文章于 2024-06-20 14:37:31 发布

阅读量2.6k

点赞数

分类专栏： opencv

本文链接：https://blog.csdn.net/mengxiang2425/article/details/82559759

版权

opencv 专栏收录该内容

16 篇文章 3 订阅

订阅专栏

融合了前面几个步骤的程序，黑白的和彩色的视差图都有了，画布的程序参考《学习opencv》这本书的，效果如下：（具体代码意思慢慢看程序理解）

#include "opencv2/opencv.hpp"
#include <highgui.h>
#include <cv.h>
#include <cxcore.h>
#include <iostream>

using namespace std;
using namespace cv;

int main()
{
    VideoCapture inputVideo1(0);
    VideoCapture inputVideo2(1);

   Mat img1;
   Mat img2;
   Mat imgCalibration1;
   Mat imgCalibration2;

   inputVideo1 >> img1;
   inputVideo2 >> img2;
   Mat cameraMatrixLeft = Mat::eye(3, 3, CV_64F);//左摄像头畸变参数（我就说怎么少了两个东西，原来双目标定的平移向量和旋转向量没求出来- -）
   cameraMatrixLeft.at<double>(0, 0) = 7.97747742e+02;
   cameraMatrixLeft.at<double>(0, 1) = 0;
   cameraMatrixLeft.at<double>(0, 2) = 3.38423920e+02;
   cameraMatrixLeft.at<double>(1, 1) = 7.95846191e+02;
   cameraMatrixLeft.at<double>(1, 2) = 2.38578918e+02;

   Mat cameraMatrixRight = Mat::eye(3, 3, CV_64F);//右摄像头畸变参数
   cameraMatrixRight.at<double>(0, 0) = 8.03603271e+02;
   cameraMatrixRight.at<double>(0, 1) = 0;
   cameraMatrixRight.at<double>(0, 2) = 3.31736206e+02;
   cameraMatrixRight.at<double>(1, 1) = 8.01632874e+02;
   cameraMatrixRight.at<double>(1, 2) = 2.66235107e+02;

   Mat distCoeffsLeft = Mat::zeros(5, 1, CV_64F);
   distCoeffsLeft.at<double>(0, 0) = -1.19046479e-01;
   distCoeffsLeft.at<double>(1, 0) = -3.16432804e-01;
   distCoeffsLeft.at<double>(2, 0) = 1.25176529e-03;
   distCoeffsLeft.at<double>(3, 0) = -1.88818539e-03;
   distCoeffsLeft.at<double>(4, 0) = 0;

   Mat distCoeffsRight = Mat::zeros(5, 1, CV_64F);
   distCoeffsRight.at<double>(0, 0) = -6.44572973e-02;
   distCoeffsRight.at<double>(1, 0) = -1.03351343e+00;
   distCoeffsRight.at<double>(2, 0) = -3.31324612e-04;
   distCoeffsRight.at<double>(3, 0) = -6.24597212e-03;
   distCoeffsRight.at<double>(4, 0) = 0;

   Mat view1, rview1, map1, map2;
   Mat view2, rview2, map3, map4;
   Rect validROIL,validROIR;//validROI是指裁剪之后的区域
   Size imageSize1;
   Size imageSize2;
   imageSize1 = img1.size();
   imageSize2 = img2.size();
   initUndistortRectifyMap(cameraMatrixLeft, distCoeffsLeft, Mat(),
       getOptimalNewCameraMatrix(cameraMatrixLeft, distCoeffsLeft, imageSize1, 1, imageSize1, 0),
       imageSize1, CV_16SC2, map1, map2);
   initUndistortRectifyMap(cameraMatrixRight, distCoeffsRight, Mat(),
       getOptimalNewCameraMatrix(cameraMatrixRight, distCoeffsRight, imageSize2, 1, imageSize2, 0),
       imageSize2, CV_16SC2, map3, map4);

   while(1)
   {
     inputVideo1 >> img1;
     inputVideo2 >> img2;

    if (img1.empty()||img2.empty())
         break;
    remap(img1, imgCalibration1, map1, map2, INTER_LINEAR);
    remap(img2, imgCalibration2, map3, map4, INTER_LINEAR);

//    imshow("left", img1);
//    imshow("right", img2);
//    imshow("Calibration1", imgCalibration1);
//    imshow("Calibration2", imgCalibration2);

    cv::StereoSGBM sgbm;//立体匹配算法
    int SADWindowSize = 11;
    sgbm.preFilterCap = 63;
    int cn=1;
    sgbm.SADWindowSize = SADWindowSize > 0 ? SADWindowSize : 3;
    sgbm.P1 = 4*cn*sgbm.SADWindowSize*sgbm.SADWindowSize;
    sgbm.P2 = 32*cn*sgbm.SADWindowSize*sgbm.SADWindowSize;
    sgbm.minDisparity = 0;
    sgbm.numberOfDisparities = 32;
    sgbm.uniquenessRatio = 10;
    sgbm.speckleWindowSize = 100;
    sgbm.speckleRange = 32;
    sgbm.disp12MaxDiff = 1;

    Mat disp, img;
    int64 t = getTickCount();

    sgbm((Mat)imgCalibration1, (Mat)imgCalibration2, disp);
    disp.convertTo(img, CV_8U, 255/(32*16.));

    imshow("disparity", img);

    Mat img_color(img.rows, img.cols, CV_8UC3);//构造RGB图像
    #define IMG_B(img,y,x) img.at<Vec3b>(y,x)[0]
    #define IMG_G(img,y,x) img.at<Vec3b>(y,x)[1]
    #define IMG_R(img,y,x) img.at<Vec3b>(y,x)[2]
    uchar tmp2=0;
    for (int y=0;y<img.rows;y++)//转为彩虹图的具体算法，主要思路是把灰度图对应的0～255的数值分别转换成彩虹色：红、橙、黄、绿、青、蓝。
     {
            for (int x=0;x<img.cols;x++)
            {
                   tmp2 = img.at<uchar>(y,x);
                   if (tmp2 <= 51)
                   {
                          IMG_B(img_color,y,x) = 255;
                          IMG_G(img_color,y,x) = tmp2*5;
                          IMG_R(img_color,y,x) = 0;
                   }
                   else if (tmp2 <= 102)
                   {
                          tmp2-=51;
                          IMG_B(img_color,y,x) = 255-tmp2*5;
                          IMG_G(img_color,y,x) = 255;
                          IMG_R(img_color,y,x) = 0;
                    }
                    else if (tmp2 <= 153)
                    {
                           tmp2-=102;
                           IMG_B(img_color,y,x) = 0;
                           IMG_G(img_color,y,x) = 255;
                           IMG_R(img_color,y,x) = tmp2*5;
                    }
                   else if (tmp2 <= 204)
                    {
                            tmp2-=153;
                            IMG_B(img_color,y,x) = 0;
                            IMG_G(img_color,y,x) = 255-uchar(128.0*tmp2/51.0+0.5);
                            IMG_R(img_color,y,x) = 255;
                    }
                    else
                     {
                            tmp2-=204;
                            IMG_B(img_color,y,x) = 0;
                            IMG_G(img_color,y,x) = 127-uchar(127.0*tmp2/51.0+0.5);
                            IMG_R(img_color,y,x) = 255;
                      }
               }
         }
    namedWindow("img_ rainbowcolor");
    imshow("img_ rainbowcolor", img_color);

    Mat canvas;//画布程序
    double sf;
    int w,h;
    Size imageSize;
    imageSize = img.size();
    sf=600./MAX(imageSize.width,imageSize.height);
    w=cvRound(imageSize.width*sf);
    h=cvRound(imageSize.height*sf);
    canvas.create(h,w*2,CV_8UC3);

    Mat canvasPart=canvas(Rect(w*0,0,w,h));
    resize(imgCalibration1,canvasPart,canvasPart.size(),0,0,INTER_AREA);
    Rect vroiL(cvRound(validROIL.x*sf), cvRound(validROIL.y*sf),                //获得被截取的区域
    cvRound(validROIL.width*sf), cvRound(validROIL.height*sf));
    rectangle(canvasPart, vroiL, Scalar(0, 0, 255), 3, 8);                      //画上一个矩形

    canvasPart=canvas(Rect(w,0,w,h));
    resize(imgCalibration2,canvasPart,canvasPart.size(),0,0,INTER_LINEAR);
    Rect vroiR(cvRound(validROIR.x*sf),cvRound(validROIR.y*sf),
               cvRound(validROIR.width*sf),cvRound(validROIR.height*sf));
    rectangle(canvasPart,vroiR,Scalar(0,255,0),3,8);

    for(int i=0;i<canvas.rows;i+=16)
        line(canvas,Point(0,i),Point(canvas.cols,i),Scalar(0,255,0),1,8);

    imshow("rectified",canvas);

    char key = waitKey(1);
    if (key == 27 || key == 'q' || key == 'Q')
         break;
  }
 cvDestroyAllWindows();

 return 0;
}