Question

我正在研究一个机器人项目是我需要检测关键点获取这些点的坐标并将它们传递给机器人来操纵。我进行了相机校准并使用校准信息来解码来自相机的图像，然后应用ORB技术来检测关键点。到目前为止一切都很好，但我不知道如何映射这些关键点坐标以与机器人一起工作。

以下是我正在使用的代码：

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

std::vector<KeyPoint> kp;
Mat frame;
Mat frame_un;
Mat camera_matrix =(Mat_<double>(3,3) << 7.4191833420713715e+02, 0.0, 320.0, 0.0, 7.4191833420713715e+02, 240.0, 0.0, 0.0,
    1.0);
Mat distortion_coefficients =(Mat_<double>(5,1) << -1.5271566741564191e-01, 1.5488166759164064e+00, 0.0, 0.0,
    -7.6517765981508861e+00);

VideoCapture cap(1); // open the default camera
if (!cap.isOpened())  // check if we succeeded
    return -1;

for (;;)
{

    cap.read(frame); // get a new frame from camera
    if (frame.empty()) 
continue;
undistort(frame, frame_un, camera_matrix, distortion_coefficients);
 // Default parameters of ORB
    int nfeatures=30;
    float scaleFactor=1.2f;
    int nlevels=8;
    int edgeThreshold=31; // Changed default (31);
    int firstLevel=0;
    int WTA_K=2;
    int scoreType=ORB::HARRIS_SCORE;
    int patchSize=31;
    int fastThreshold=20;

Ptr<ORB> detector = ORB::create(
    nfeatures,
    scaleFactor,
    nlevels,
    edgeThreshold,
    firstLevel,
    WTA_K,
    scoreType,
    patchSize,
    fastThreshold );

    detector->detect(frame_un, kp);
    std::cout << "Found " << kp.size() << " Keypoints " << std::endl;
for(int i=0; i<=kp.size();i++)
{ 
int x = kp[i].pt.x;
int y = kp[i].pt.y;
cout << "Point "<<i<<" Xpos = " << x <<  " Point "<<i<< " Ypos = " << y << "\n";

}    
Mat out;

    //drawKeypoints(img, kp, out, Scalar::all(255));
drawKeypoints(frame_un, kp, out, Scalar::all (255));
namedWindow("Kpts", WINDOW_FREERATIO);
    imshow("Kpts", out);
waitKey(0);
destroyWindow("Kpts");
}
   // waitKey(0);
    return 0;
}

Answer 1

只需将坐标系映射到机器人的基本坐标系并将其作为原点。在此之前确保您已经准确地校准了您的视力。

如何将像素坐标转换为世界坐标？

1 个答案: