毕设:opencv3.4.2 c++ 相机标定

网上全是opencv2的 用也用不了

而且路径什么的很重要 要改成自己的哈!!

读了两次图片calibdata.txt 用的都是jpg 不是bmp

 

 

#include
#include
#include "opencv2/opencv.hpp"
#include
#include
#include

using namespace std;

int main()
{
    ifstream fin("calibdata.txt"); /* 标定所用图像文件的路径 */
    ofstream fout("caliberation_result.txt");  /* 保存标定结果的文件 */

    if (!fin){
        std::cout << "Calibration image txt read failed" << endl;
        return 0;
    }

    //读取每一幅图像，从中提取出角点，然后对角点进行亚像素精确化
    std::cout << "开始提取角点………………";
    int image_count = 0;  /* 图像数量 */
    cv::Size image_size;  /* 图像的尺寸 */
    cv::Size board_size = cv::Size(5,7);    /* 标定板上每行、列的角点数 */
    vector image_points_buf;  /* 缓存每幅图像上检测到的角点 */
    vector > image_points_seq; /* 保存检测到的所有角点 */
    string filename;

    int count = -1;//用于存储角点个数。
    while (getline(fin, filename))
    {
        image_count++;
        // 用于观察检验输出
        std::cout << "image_count = " << image_count << endl;
        /* 输出检验*/
        std::cout << "-->count = " << count;
        cv::Mat dst = cv::imread(filename);
        //cv::Mat imageInput;
        //cv::Size dsize = cv::Size(round(0.3 * dst.cols), round(0.3 * dst.rows));
        //cv::resize(dst,imageInput,dsize,0,0,CV_INTER_AREA);
        if (image_count == 1)  //读入第一张图片时获取图像宽高信息
        {
            image_size.width = dst.cols;
            image_size.height = dst.rows;
            std::cout << "image_size.width = " << image_size.width << endl;
            std::cout << "image_size.height = " << image_size.height << endl;

        }

        /* 提取角点 */
        if (0 == cv::findChessboardCorners(dst, board_size, image_points_buf))
        {
            std::cout << "can not find chessboard corners!\n"; //找不到角点
            exit(1);
        }
        else
        {
            cv::Mat view_gray;
            cv::cvtColor(dst, view_gray, CV_RGB2GRAY);
            /* 亚像素精确化 */
            cv::find4QuadCornerSubpix(view_gray, image_points_buf, cv::Size(5, 5)); //对粗提取的角点进行精确化
            //cornerSubPix(view_gray,image_points_buf,Size(5,5),Size(-1,-1),TermCriteria(CV_TERMCRIT_EPS+CV_TERMCRIT_ITER,30,0.1));
            image_points_seq.push_back(image_points_buf);  //保存亚像素角点
            /* 在图像上显示角点位置 */
            cv::drawChessboardCorners(view_gray, board_size, image_points_buf, false); //用于在图片中标记角点
            imshow("Camera Calibration", view_gray);//显示图片
            cv::waitKey(500);//暂停0.5S
        }
    }

    int total = image_points_seq.size();
    std::cout << "total = " << total << endl;
    int CornerNum = board_size.width*board_size.height;  //每张图片上总的角点数
    for (int ii = 0; ii     {
        if (0 == ii%CornerNum)// 24 是每幅图片的角点个数。此判断语句是为了输出 图片号，便于控制台观看
        {
            int i = -1;
            i = ii / CornerNum;
            int j = i + 1;
            std::cout << "--> 第 " << j << "图片的数据 --> : " << endl;
        }
        if (0 == ii % 3)    // 此判断语句，格式化输出，便于控制台查看
        {
            std::cout << endl;
        }
        else
        {
            std::cout.width(10);
        }
        //输出所有的角点
        std::cout << " -->" << image_points_seq[ii][0].x;
        std::cout << " -->" << image_points_seq[ii][0].y;
    }
    std::cout << "角点提取完成！\n";

    //以下是摄像机标定
    std::cout << "开始标定………………";
    /*棋盘三维信息*/
    cv::Size square_size = cv::Size(10, 10);  /* 实际测量得到的标定板上每个棋盘格的大小 */
    vector > object_points; /* 保存标定板上角点的三维坐标 */
    /*内外参数*/
    //cv::Mat()
    cv::Mat cameraMatrix = cv::Mat(3, 3, CV_32FC1, cv::Scalar::all(0)); /* 摄像机内参数矩阵 */
    vector point_counts;  // 每幅图像中角点的数量
    cv::Mat distCoeffs = cv::Mat(1, 5, CV_32FC1, cv::Scalar::all(0)); /* 摄像机的5个畸变系数：k1,k2,p1,p2,k3 */
    vector tvecsMat;  /* 每幅图像的旋转向量 */
    vector rvecsMat; /* 每幅图像的平移向量 */
    /* 初始化标定板上角点的三维坐标 */
    int i, j, t;
    for (t = 0; t     {
        vector tempPointSet;
        for (i = 0; i         {
            for (j = 0; j             {
                cv::Point3f realPoint;
                /* 假设标定板放在世界坐标系中z=0的平面上 */
                realPoint.x = i*square_size.width;
                realPoint.y = j*square_size.height;
                realPoint.z = 0;
                tempPointSet.push_back(realPoint);
            }
        }
        object_points.push_back(tempPointSet);
    }
    /* 初始化每幅图像中的角点数量，假定每幅图像中都可以看到完整的标定板 */
    for (i = 0; i     {
        point_counts.push_back(board_size.width*board_size.height);
    }
    /* 开始标定 */
    calibrateCamera(object_points, image_points_seq, image_size, cameraMatrix, distCoeffs, rvecsMat, tvecsMat, 0);
    cout << "标定完成！\n";
    //对标定结果进行评价
    cout << "开始评价标定结果………………\n";
    double total_err = 0.0; /* 所有图像的平均误差的总和 */
    double err = 0.0; /* 每幅图像的平均误差 */
    vector image_points2; /* 保存重新计算得到的投影点 */
    cout << "\t每幅图像的标定误差：\n";
    fout << "每幅图像的标定误差：\n";
    for (i = 0; i     {
        vector tempPointSet = object_points[i];
        /* 通过得到的摄像机内外参数，对空间的三维点进行重新投影计算，得到新的投影点 */
        projectPoints(tempPointSet, rvecsMat[i], tvecsMat[i], cameraMatrix, distCoeffs, image_points2);
        /* 计算新的投影点和旧的投影点之间的误差*/
        vector tempImagePoint = image_points_seq[i];
        cv::Mat tempImagePointMat = cv::Mat(1, tempImagePoint.size(), CV_32FC2);
        cv::Mat image_points2Mat = cv::Mat(1, image_points2.size(), CV_32FC2);
        for (int j = 0; j < tempImagePoint.size(); j++)
        {
            image_points2Mat.at(0, j) = cv::Vec2f(image_points2[j].x, image_points2[j].y);
            tempImagePointMat.at(0, j) = cv::Vec2f(tempImagePoint[j].x, tempImagePoint[j].y);
        }
        err = norm(image_points2Mat, tempImagePointMat, cv::NORM_L2);
        total_err += err /= point_counts[i];
        std::cout << "第" << i + 1 << "幅图像的平均误差：" << err << "像素" << endl;
        fout << "第" << i + 1 << "幅图像的平均误差：" << err << "像素" << endl;
    }
    std::cout << "总体平均误差：" << total_err / image_count << "像素" << endl;
    fout << "总体平均误差：" << total_err / image_count << "像素" << endl << endl;
    std::cout << "评价完成！" << endl;
    //保存定标结果    
    std::cout << "开始保存定标结果………………" << endl;
    cv::Mat rotation_matrix = cv::Mat(3, 3, CV_32FC1, cv::Scalar::all(0)); /* 保存每幅图像的旋转矩阵 */
    fout << "相机内参数矩阵：" << endl;
    fout << cameraMatrix << endl << endl;
    fout << "畸变系数：\n";
    fout << distCoeffs << endl << endl << endl;
    for (int i = 0; i     {
        fout << "第" << i + 1 << "幅图像的旋转向量：" << endl;
        fout << tvecsMat[i] << endl;
        /* 将旋转向量转换为相对应的旋转矩阵 */
        Rodrigues(tvecsMat[i], rotation_matrix);
        fout << "第" << i + 1 << "幅图像的旋转矩阵：" << endl;
        fout << rotation_matrix << endl;
        fout << "第" << i + 1 << "幅图像的平移向量：" << endl;
        fout << rvecsMat[i] << endl << endl;
    }
    std::cout << "完成保存" << endl;
    fout << endl;
    /************************************************************************
    显示定标结果
    *************************************************************************/
    cv::Mat mapx = cv::Mat(image_size, CV_32FC1);
    cv::Mat mapy = cv::Mat(image_size, CV_32FC1);
    cv::Mat R = cv::Mat::eye(3, 3, CV_32F);
    std::cout << "保存矫正图像" << endl;
    string imageFileName;
    std::stringstream StrStm;
    for (int i = 0; i != image_count; i++)
    {
        std::cout << "Frame #" << i + 1 << "..." << endl;
        cv::initUndistortRectifyMap(cameraMatrix, distCoeffs, R, cameraMatrix, image_size, CV_32FC1, mapx, mapy);
        StrStm.clear();
        imageFileName.clear();
        string filePath = "/home/kong/opencv_cam_calibration/";
        StrStm << i + 1;
        StrStm >> imageFileName;
        filePath += imageFileName;
        filePath += ".jpg";
        cv::Mat imageSource = cv::imread(filePath);
        cv::Mat newimage = imageSource.clone();

        if(imageSource.empty())
            cout<<"imageSource is null"<         //另一种不需要转换矩阵的方式
        //cv::undistort(imageSource,newimage,cameraMatrix,distCoeffs);
        cv::remap(imageSource, newimage, mapx, mapy, cv::INTER_LINEAR,cv::BORDER_CONSTANT, cv::Scalar(0, 0, 0));
        std::cout<<"successful2  "<         StrStm.clear();
        filePath.clear();
        StrStm << i + 1;
        StrStm >> imageFileName;
        imageFileName += "_d.jpg";
        cv::imwrite(imageFileName, newimage);
    }
    std::cout << "保存结束" << endl;

    return 0 ;
}