somedev's profile - activity

the solvePnP() method takes the following as input :

vector<point3f> ObjectPoints, vector<point2f> imagePoints, Mat cameraMatrix, Mat distortionCoeffs, Mat rvec, Mat tvec

The parameters you seem to be passing are:

vector<vector<point3f> > boardPoints, vector<point2f> boardCorners, vector<mat> intrinsics, vector<mat> distortion, vector<mat> rvec, vector<mat> tvec

solvePnP() doesn't implicitly convert vector<vector<t> > to vector<t> like the calibrateCamera() method does, try to pass the parameters iteratively, it will work. Someone should include this in the documentation to avoid confusion.

To make the cameraCalibrate function use the initial cameraMatrix you specified, you have to give a flags parameter with the value CV_CALIB_USE_INTRINSIC_GUESS;

try

calibrateCamera(objectPoints,imagePoints,imageSize,cameraMatrix,distCoeffs,rvecs,tvecs,CV_CALIB_USE_INTRINSIC_GUESS);