Opencv 2 Mat- Error on runtime

Question

I am starting with the basics of image processing and I am trying to convert a RGB image into grey scale using averaging.

My code is

#include<opencv2/core/core.hpp> #include<opencv2/highgui/highgui.hpp> #include<opencv2/imgproc/imgproc.hpp> #include<iostream> using namespace cv; int main() { Mat image = imread("a.jpg"); namedWindow("Original "); namedWindow("Grey"); imshow("Original", image); Mat grey; std::cout << "hello\n"; for (int i = 0; i < image.rows; i++) { for (int j = 0; j < image.cols; j++) { grey.at<Vec3b>(i, j) = (image.at<Vec3b>(i, j)[0] + image.at<Vec3b>(i, j)[1] + image.at<Vec3b>(i, j)[2]); } } imshow("Grey", grey); return 0; }`

I think there is some problem with the Mat Grey , while accessing the element inside the for loop.

scap3y · Accepted Answer · 2014-05-21 09:36:06Z

2

You need to initialize your grey,

Mat grey(image.rows, image.cols, CV_8UC1);

And then, during processing,

 for (int j = 0; j < image.cols; j++) { grey.at<Vec3b>(i, j) = ((image.at<Vec3b>(i, j)[0] + image.at<Vec3b>(i, j)[1] + image.at<Vec3b>(i, j)[2]))/3; }

The /3 will give you the true grayscale value.

HTH

answered May 21, 2014 at 9:36

scap3y

1,19810 silver badges27 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

scap3y Over a year ago

If you are talking about weighted division, then that isn't true grayscale; that is just considering the colors which have the highest response in the human vision system.

thedarkside ofthemoon Over a year ago

Can you please tell me where is the definition of your "true grayscale"?

thedarkside ofthemoon Over a year ago

Every grayscale transformation is done by Y = 0.2126*R + 0.7152*G + 0.0722*B. And by the way, OpenCV loads the images in BRG mode.

scap3y Over a year ago

This definition is indeed correct for Grayscale conversion based on human perception since cones are most sensitive to Green. If, however, a true conversion it to be applied, the weights should be distributed equally. This would ensure that a perception system has the best response to all the colors and not just one.

thedarkside ofthemoon Over a year ago

so you are saying that your grayscale is true because is not human? Strange... Anyway, Can you give a link that sustains your idea?

|

thedarkside ofthemoon · Accepted Answer · 2014-05-21 09:41:16Z

Of course, You have not initialized it, so it is empty, you cannot access the elements.

Mat grey = Mat::zeros(image.rows, image.cols, CV_8UC1); Should fix it.

Try using an if statement before accessing the elements of a Mat:

if (grey.empty()) { std::cout << "Empty Mat!\n"; return -1; }

Another thing is that adding the image channels are not the true grayacale, see this.

Collectives™ on Stack Overflow

Opencv 2 Mat- Error on runtime

2 Answers 2

6 Comments

Comments

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

6 Comments

Comments

Related