CODEWITHSUNDEEP
c++qtopencv
user1003131
user1003131

Reputation: 537

OpenCV corner detection returning null corners vector

I am trying to detect the corners of a chessboard image. I have tried using goodFeaturesToTrack and findChessboardCorners but they are both failing in the same way. They corner vector returned seems to be null. (i.e. size is -2948392...memory error type of stuff). 

  gray = imread("C:/Users/Wesley/Desktop/New folder/cb.png",0);

  // Parameters for Shi-Tomasi algorithm
  std::vector<cv::Point2f> corners;
  double qualityLevel = 0.1;
  double minDistance = 10;
  int blockSize = 3;
  bool useHarrisDetector = false;
  double k = 0.04;
  int maxCorners = 30;
  Mat copy;
  copy = gray.clone();

  // Apply corner detection
  goodFeaturesToTrack(gray,
                 corners,
                 maxCorners,
                 qualityLevel,
                 minDistance,
                 Mat(),
                 blockSize,
                 useHarrisDetector,
                 k );

      int r = 4;
      for( int i = 0; i < corners.size(); i++ )
      { circle( copy, corners[i], r, Scalar(255, 255,255), -1, 8, 0 ); }

  // Show what you got
  imshow( "source_window", copy );

It must be something I am doing wrong. It crashes as soon as corners.size() is called due to a memory access violation. This is in Qt if it makes any difference.

Hopefully someone can point me in the right direction!

Upvotes: 0

Views: 339

Answers (1)

Micka
Micka

Reputation: 20160

Please check that you aren't compiling in release mode but linking against debug mode or vice versa.

OpenCV gives you the strangest behaviour if you link against the wrong libraries.

Upvotes: 2

Related Questions