我首先使用 opencv 包 (stereo_calib.cpp) 中的示例代码校准了我的相机。现在我想用它来纠正立体匹配的图像。我开始纠正一个图像，但我无法让它工作，因为 initUndistortRectifyMap 返回一个不包含任何内容的地图。我无法找出我做错了什么。
Mat ciml, cimr, iml, imr, imlCalibrated, imrCalibrated;
vector<KeyPoint> keyl, keyr, keyInPlane, keyMatched;
//-- Load Images and convert to gray scale
string filenamel = "img/imL2.png";
string filenamer = "img/imR2.png";
ciml = imread(filenamel, 0);
cvtColor(iml, ciml, CV_GRAY2BGR);
cimr = imread(filenamer, 0);
cvtColor(imr, cimr, CV_GRAY2BGR);
Size imageSize = iml.size();
//-- Load calibration matrices
FileStorage fs_in, fs_ex;
fs_in.open( "intrinsics.yml", FileStorage::READ );
fs_ex.open( "extrinsics.yml", FileStorage::READ );
Mat rmap[2][2], cameraMatL, cameraMatR, coeffMatL, coeffMatR, R1, R2, P1, P2;
fs_ex["R1"] >> R1; fs_ex["R2"] >> R2;
fs_ex["P1"] >> P1; fs_ex["P2"] >> P2;
fs_in["M1"] >> cameraMatL; fs_in["M2"] >> cameraMatR;
fs_in["D1"] >> coeffMatL; fs_in["D2"] >> coeffMatR;
cout << "camera Matrix: " << endl << cameraMatL << endl << endl;
//-- Find remap values
initUndistortRectifyMap(cameraMatL, coeffMatL, R1, P1, imageSize, CV_16SC2, rmap[0][0], rmap[0][1]);
initUndistortRectifyMap(cameraMatR, coeffMatR, R2, P2, imageSize, CV_16SC2, rmap[1][0], rmap[1][1]);
回答：这很简单，因为图像没有加载，图像大小为零，所以地图是一个空矩阵。我会尽快发布答案。