Eren V.
Eren V.

Reputation: 210

Findcontours() run-time error

I am using a function like this;

Mat large = imread(path+name);
Mat rgb;
if (large.rows > 2500 || large.cols > 1250)
{
    pyrDown(large, rgb);
}
else
{
    rgb = large.clone();
}
cv::Mat smallx;
cvtColor(rgb, smallx, CV_BGR2GRAY);
Mat grad,connected,bw;

Mat morphKernel = getStructuringElement(MORPH_ELLIPSE, Size(3, 3));
cv::morphologyEx(smallx, grad, MORPH_GRADIENT, morphKernel);
cv::threshold(grad, bw, 100, 255, THRESH_BINARY + THRESH_OTSU);
morphKernel = getStructuringElement(MORPH_RECT, Size(9, 1));
cv::morphologyEx(bw, connected, MORPH_CLOSE, morphKernel);

Mat mask = Mat::zeros(bw.size(), CV_8UC1);
vector<vector<Point>> contours;
vector<Vec4i> hierarchy;

cv::findContours(connected, contours, hierarchy, CV_RETR_CCOMP, CV_CHAIN_APPROX_SIMPLE, Point(0, 0));

Some input images prompt exception when function returns, it throws me to "_Mybase::deallocate(_Ptr, _Count);" code block in xmemory0 class. I must catch this kind of errors and prevent this process from unexpected suspensions

--------------ERROR DETAILS--------------

I tried that code block with 30 different input but it gave error in a few image. after findcontours it thrown an exception break and show me this message.

Unhandled exception at 0x74477fb2 (ucrtbase.dll) in ImageRecognizer.exe: 0xC0000409: 0xc0000409.

in that time , "vector" class opening in my Visual Studio with these lines ;

~vector() _NOEXCEPT
    {   // destroy the object
    _Tidy();
    }

or "xmemory" class opening with these lines;

void deallocate(pointer _Ptr, size_type _Count)
    {   // deallocate object at _Ptr, ignore size
    _Mybase::deallocate(_Ptr, _Count);
    }

Upvotes: 0

Views: 371

Answers (2)

Eren V.
Eren V.

Reputation: 210

I tried to OpenCV 2.4.13 instead of OpenCV 2.4.12 and problem solved by itself. It wasn't an algorithm or logical error. It was a library bug.

Upvotes: 0

hagor
hagor

Reputation: 314

It seems that number of counters your method finds exceeds 50000. Try not to use magic numbers. It is always a bad practice.

Upvotes: 1

Related Questions