检查cv :: Point是否在cv :: Mat中

Question

Does anyone know if Opencv provide a function to check if a cv::Point is inside a cv::Mat ?

Basically I am doing:

int x = (current.x - offset);
int y = current.y;
if (x >= 0 && y >= 0 && x < mat.cols &&  y < mat.rows) && ((int)mat.at<uchar>(y, x) == 0)){
        return cv::Point(x, y);
    }
}

I would like to know if there is something faster ? Or if this was bad to do this ?

Answer 1

You can construct a cv::Rect of the size as the cv::Mat and use its contains() method:

cv::Rect rect(cv::Point(), mat.size());
cv::Point p(x, y);

if (rect.contains(p) && mat.at<uchar>(y, x) == 0)
{
  return p;
}

Alternatively, you can catch exceptions in at() if indices are out of bounds:

UPD: As properly mentioned by @Antonio in the comments, the following works only in debug mode, since " For the sake of higher performance, the index range checks are only performed in the Debug configuration ", which is kind of surprising and is different from how std::vector::at() works.

try
{
  if (mat.at<uchar>(y, x) == 0)
  {
    return cv::Point(x, y);
  }
}
catch (cv::Exception& e)
{
}

However, be aware of the potential performance penalty caused by exceptions. You shouldn't use the latter approach if this statement is executed in a loop or just very often. Or in case it is normal rather than exceptional situation.