Check if cvRect is within cvMat

Question

I am using bounding boxes to find text in an image and sometimes the boxes are just slightly too small and cut off parts of text on the top and bottom.

So I thought I'd extend every bounding box just a little to compensate for the inaccuracy.

double Z = 10;
cv::Rect extended( appRect.x-Z, appRect.y-Z, appRect.width+2*Z, appRect.height+2*Z);

appRect being a cv::Rect

This does what I am looking for however it seems that sometimes it takes the bounding box out of bounds.

Giving me this error:

OpenCV Error: Assertion failed (0 <= roi.x && 0 <= roi.width && roi.x + roi.width <= m.cols && 0 <= roi.y && 0 <= roi.height && roi.y + roi.height <= m.rows) in Mat

How can I check that the rect is within the boundaries of the image and avoid this error while still expanding the bounding boxes?

Answer 1

Like the error says, your x, y, w and h are not allowed to be negative.

Try to add a std::max() and std::min():

#include <algorithm> // std::max && std::min

int Z = 10;
int x = std::max<int>(0, appRect.x-Z);
int y = std::max<int>(0, appRect.y-Z);
int w = std::min<int>(mat.cols - x, appRect.width+2*Z);
int h = std::min<int>(mat.rows - y, appRect.height+2*Z);
cv::Rect extended(x, y, w, h);

Or as Iwillnotexist smartly suggested:

// expand
double Z = 10;
cv::Rect extended( appRect.x-Z, appRect.y-Z, appRect.width+2*Z, appRect.height+2*Z);

// intersect
extended &= Rect(Point(0, 0), mat.size());