stereoRectify makes strange images distortion

asked 2015-07-12 09:54:22 -0600

Markov Alex gravatar image

Hello. I am trying to compute point cloud from two cameras.

  1. Find chessboard corners using findChessboardCorners
  2. Calibrate cameras using stereoCalibrate
  3. Calculate rectify matrix using stereoRectify
  4. Calculate remap maps using initUndistortRectifyMap
  5. Remap images

And I got a very bad results: image description

It's not so bad like this: image description

Is the any way to fix it?

This is my code (also available on GitHub): // left - vector of grayscale images from left camera // right - vector of grayscale images from right camera // patternSize - number of chessboard corners StereoCalibData StereoVision::Calibrate(const std::vector<cv::mat>& left, const std::vector<cv::mat>& right, cv::Size patternSize) { cv::Size imSize;

// Images points
std::vector<cv::Point2f> imagePointsLeftSingle, imagePointsRightSingle;
std::vector<std::vector<cv::Point2f>> imagePointsLeft, imagePointsRight;

// Object points
std::vector<cv::Point3f> objectPointsSingle;
std::vector<std::vector<cv::Point3f>> objectPoints;

//stereoCalibrate output
cv::Mat CM1(3, 3, CV_64FC1), CM2(3, 3, CV_64FC1);   //Camera matrix
cv::Mat D1, D2;         //Distortion
cv::Mat R,              //Rotation matrix
        T,              //Translation matrix
        E,              //Существенная матрица
        F;              //Фундаментальная матрица

//stereoRectify output
cv::Mat R1, R2, P1, P2;
cv::Mat Q;

//1. ---------------------------  Find chessboard corners ---------------------------------------------------------------

for (int i = 0; i < left.size(); i++)
    //Finding chessboard corners
    bool isFoundLeft = cv::findChessboardCorners(left[i], patternSize, imagePointsLeftSingle, CV_CALIB_CB_ADAPTIVE_THRESH | CV_CALIB_CB_FILTER_QUADS);
    bool isFoundRight = cv::findChessboardCorners(right[i], patternSize, imagePointsRightSingle, CV_CALIB_CB_ADAPTIVE_THRESH | CV_CALIB_CB_FILTER_QUADS);

    imSize = left[i].size();

    cornerSubPix(left[i], imagePointsLeftSingle, cv::Size(11, 11), cv::Size(-1, -1), cv::TermCriteria(CV_TERMCRIT_EPS + CV_TERMCRIT_ITER, 30, 0.1));
    cornerSubPix(right[i], imagePointsRightSingle, cv::Size(11, 11), cv::Size(-1, -1), cv::TermCriteria(CV_TERMCRIT_EPS + CV_TERMCRIT_ITER, 30, 0.1));

    //Sometimes chessboard corners are found in bad direction (vertically lines), so I "rotate" them
    _fixChessboardCorners(imagePointsLeftSingle, patternSize);
    _fixChessboardCorners(imagePointsRightSingle, patternSize);

    //Add to the corners vectors

//2. --------------------------- Compute object points ------------------------------------------------------------------

for (int j = 0; j<patternSize.height*patternSize.width; j++)
    objectPointsSingle.push_back(cv::Point3f(j / patternSize.width, j%patternSize.width, 0.0f));

for (int i = 0; i < left.size(); i++)

//3. --------------------------- Stereo calibration --------------------------------------------------------------------------
stereoCalibrate(objectPoints, imagePointsLeft, imagePointsRight,
    cvTermCriteria(CV_TERMCRIT_ITER + CV_TERMCRIT_EPS, 100, 1e-5));

cv::stereoRectify(CM1, D1, CM2, D2, imSize, R, T, R1, R2, P1, P2, Q);

// -------------------------- Saving rsults ------------------------------------------------------------------------

calibData.ImageSize = imSize;
calibData.LeftCameraMatrix = CM1.clone();
calibData.RightCameraMatrix = CM2.clone();
calibData.LeftCameraDistortions = D1.clone();
calibData.RightCameraDistortions = D2.clone();
calibData.LeftCameraRectifiedProjection = P1.clone();
calibData.RightCameraRectifiedProjection = P2.clone();
calibData.LeftCameraRot = R1.clone();
calibData.RightCameraRot = R2.clone();
calibData.Q = Q;

return calibData;


IPointCloudStorage* StereoVision::CalculatePointCloud(const cv::Mat& left, const cv::Mat& right) const { cv::Mat leftRemaped, rightRemaped; cv::Mat depth, normalDepth;

cv::initUndistortRectifyMap(data.LeftCameraMatrix, data.LeftCameraDistortions, data.LeftCameraRot, data.LeftCameraRectifiedProjection, data.ImageSize, CV_32FC1, data.LeftMapX, data.LeftMapY);
cv::initUndistortRectifyMap(data.RightCameraMatrix, data.RightCameraDistortions, data.RightCameraRot, data.RightCameraRectifiedProjection, data.ImageSize, CV_32FC1, data.RightMapX, data.RightMapY);

cv::remap(left, leftRemaped, calibData.LeftMapX, calibData.LeftMapY, cv::INTER_LINEAR, cv::BORDER_CONSTANT, cv::Scalar());
cv::remap(right, rightRemaped, calibData.RightMapX, calibData.RightMapY, cv::INTER_LINEAR, cv::BORDER_CONSTANT, cv::Scalar());

cv::imshow("w1", leftRemaped);
cv::imshow("w2", rightRemaped);
return NULL;


P.S. There are some problems with syntax highlighting

edit retag flag offensive close merge delete


  • please use a rectangular chessboard, not a quadratic one. (you can't resolve 90° rotation with the quadratic one)
  • you will need about a dozen good chessboard pairs for a successful calibration. check your rms (the return value). if it is above 0.5, you'll have to start again.
  • "P.S. There are some problems with syntax highlighting" - edit, mark your code, and press '10101' button for proper formatting
berak gravatar imageberak ( 2015-07-12 10:09:47 -0600 )edit

you will need about a dozen good chessboard pairs for a successful calibration You mean that I need do many photos of chessboard with different poses and angles?

Markov Alex gravatar imageMarkov Alex ( 2015-07-12 10:42:02 -0600 )edit

many poses of the same chessboard will do. but yes, different angles and poses. try to cover the whole visible area with multiple shots

berak gravatar imageberak ( 2015-07-12 10:46:19 -0600 )edit

I am trying with rectangle chessboard, do multiple shots, but result is so bad. The most frqfrequent result is just two black images.

Markov Alex gravatar imageMarkov Alex ( 2015-07-12 11:20:55 -0600 )edit