如何进行线结构光与单目相机的联合标定

实验室同学一起劳动的成果，后期进行理论说明

#include <iostream>
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/features2d/features2d.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/calib3d/calib3d.hpp>
//#include <opencv2/nonfree/nonfree.hpp>


#include <pcl/io/pcd_io.h>
#include <pcl/point_types.h>
#include <pcl/visualization/cloud_viewer.h>
#include <pcl/filters/radius_outlier_removal.h>
#include <pcl/filters/conditional_removal.h>
#include <pcl/filters/statistical_outlier_removal.h>
#include <pcl/ModelCoefficients.h>
#include <pcl/point_types.h>
#include <pcl/sample_consensus/method_types.h>
#include <pcl/sample_consensus/model_types.h>
#include <pcl/segmentation/sac_segmentation.h>


#include <PeakFinder.h>
#include <math.h>       /* sin */


#include "pcl_ros/point_cloud.h"


using namespace std;
using namespace cv;


typedef pcl::PointXYZRGB PointT;
typedef pcl::PointCloud<PointT> PointCloud;


pcl::visualization::CloudViewer viewer ("Simple Cloud Viewer");


#define fx 1204.696659
#define fy 1204.299368
#define cx 966.428218
#define cy 559.772423
#define image_width 1920
#define image_height 1080
#define theta 0.00    //in radians
#define baseline 0.1000


int main(int argc, char** argv)
{
    cv::Mat black;
    cv::Mat frame;
    cv::Mat temp;
    cv::Mat diff;


    cv::Mat imageGrayblack;
    cv::Mat imageGray;


    cv::Mat imgundistortedblack;
    cv::Mat imgundistorted;


    PointCloud::Ptr cloud ( new PointCloud );
    vector<Point3f> point3d;
    double scale=100000.0;






    float col[image_width/2]={0.0};
    soundtouch::PeakFinder pf;
    double pp=0.0;           //accurate peak postion


    cv::Matx33f K (fx,0,cx,
                   0,fy,cy,
                   0,0,1);    //calibration matrix


    vector<float> distCoeffs = {-0.400622, 0.134591, 0.001135, -0.001642, 0.000000};//c++ 11 in CMakeLists.txt!!!
    cv::Size patternsize(11,8); //interior number of corners
    vector<Point2f> corners; //this will be filled by the detected corners
    bool patternfound=false;
    int num=270;//0.25*1080
    vector<double> vpp(image_height,0.0);


    for(int kk=0;kk<3;kk++)
    {
        if(kk==0)
        {
            frame=cv::imread("/home/kylefan/Pictures/Webcam/2017-03-10-140619.jpg");
            black=cv::imread("/home/kylefan/Pictures/Webcam/2017-03-10-140626.jpg");
        }
        if(kk==1)
        {
            frame=cv::imread("/home/kylefan/Pictures/Webcam/2017-03-10-141006.jpg");
            black=cv::imread("/home/kylefan/Pictures/Webcam/2017-03-10-141014.jpg");
        }
        if(kk==2)
        {
            frame=cv::imread("/home/kylefan/Pictures/Webcam/2017-03-10-140913.jpg");
            black=cv::imread("/home/kylefan/Pictures/Webcam/2017-03-10-140921.jpg");
        }


        cvtColor(frame, imageGray, CV_BGR2GRAY);
        cvtColor(black, imageGrayblack, CV_BGR2GRAY);


        cv::undistort(imageGray,imgundistorted,K, distCoeffs, K );
        cv::undistort(imageGrayblack,imgundistortedblack,K, distCoeffs, K );


        cv::absdiff(imgundistorted,imgundistortedblack,diff);


        cv::GaussianBlur(imgundistorted,imgundistorted,cv::Size(5,5),2,2);


    //1.find corners in image without stripe
        //CALIB_CB_FAST_CHECK saves a lot of time on images
        //that do not contain any chessboard corners
        patternfound = findChessboardCorners(imgundistortedblack, patternsize, corners,CALIB_CB_ADAPTIVE_THRESH);
        cout<<"patternfound:"<<patternfound<<endl;
        if(patternfound)cornerSubPix(imgundistortedblack, corners, Size(11, 11), Size(-1, -1),TermCriteria(CV_TERMCRIT_EPS + CV_TERMCRIT_ITER, 30, 0.1));
        drawChessboardCorners(imgundistortedblack, patternsize, Mat(corners), patternfound);
        //for(int i=0;i<corners.size();i++)cout<<"("<<corners[i].x<<","<<corners[i].y<<")"<<endl;


    //2,calculate 3-d of corners in camera coordinate.remember :2d corners must be in the undistorted images !!!!!!!  .but why?
        for(int i=0;i<corners.size();i++)
        {
            Point3f p3d;
            //cv::Matx31f point (corners[i].x-image_width/2,image_height/2-corners[i].y,1);
            cv::Matx31f point (corners[i].x,corners[i].y,1);
            cv::Mat_<double> point3=cv::Mat(K.inv())*cv::Mat(point);


            //cout<<"K.inv()="<<K.inv()<<endl;
            p3d.x=point3.at<double>(0,0);
            p3d.y=point3.at<double>(1,0);
            p3d.z=point3.at<double>(2,0);
            point3d.push_back(p3d);
        }


    //3,calculate scale
        double distance = cv::norm(point3d[0]-point3d[10]);
        //cout<<"distance="<<distance<<endl;
        scale=distance/0.200;


        for(int i=0;i<corners.size();i++)
        {
            point3d[i].x/=scale;
            point3d[i].y/=scale;
            point3d[i].z/=scale;
            //cout<<point3d[i].x<<","<<point3d[i].y<<","<<point3d[i].z<<endl;
        }




    //4,get stripe point


        for(int i=num;i<2*num;i++)
        {
            for(int j=image_width/2;j<image_width;j++)
            {
                col[j-image_width/2]=(float)(diff.at<uchar>(i,j));
            }
            pp=pf.detectPeak(col,0,image_width/2);
            diff.at<uchar>(i,(int)pp+image_width/2)=0;
            PointT p;
            //cv::Matx31f point (pp+(double)image_width/2.0-image_width/2,image_height/2-(i+num),1);
            cv::Matx31f point (pp+(double)image_width/2.0,i+num,1);
            cv::Mat_<double> point3=cv::Mat(K.inv())*cv::Mat(point);
            p.x=point3.at<double>(0,0)/scale;
            p.y=point3.at<double>(1,0)/scale;
            p.z=point3.at<double>(2,0)/scale;


            //cout<<p.x<<","<<p.y<<","<<p.z<<endl;


            p.r=55;
            p.g=250;
            p.b=55;
            cloud->push_back(p);
        }
        cout<<"cloud num="<<cloud->width<<endl;
        point3d.clear();
        corners.clear();
        patternfound=false;
        scale=100000.0;


    }


    pcl::PCLHeader header;
    header.frame_id="map";
    cloud->header=header;
    cloud->height = 1;
    cloud->width = cloud->points.size();
    cloud->is_dense = false;
    cloud->points.resize (cloud->width * cloud->height);


    pcl::ModelCoefficients::Ptr coefficients (new pcl::ModelCoefficients);
    pcl::PointIndices::Ptr inliers (new pcl::PointIndices);
    // Create the segmentation object
    pcl::SACSegmentation<PointT> seg;
    // Optional
    seg.setOptimizeCoefficients (true);
    // Mandatory
    seg.setModelType (pcl::SACMODEL_PLANE);
    seg.setMethodType (pcl::SAC_RANSAC);
    seg.setDistanceThreshold (0.01);


    seg.setInputCloud (cloud);
    seg.segment (*inliers, *coefficients);


    if (inliers->indices.size () == 0)
    {
      PCL_ERROR ("Could not estimate a planar model for the given dataset.");
      return (-1);
    }


    std::cerr << "Plane : " <<coefficients->values[0] << "x+"<< coefficients->values[1] << "y+"<< coefficients->values[2] << "z+"<< coefficients->values[3] <<"=0"<< std::endl;
    viewer.showCloud(cloud);
    while (!viewer.wasStopped ());
}

http://blog.csdn.net/fk1174/article/details/61201231