Create a Mat from coordinates stored in another Mat

edit

So, basically in python y have the following

img[np.array(coor[:, 1]), np.array(coor[:, 0])]

How can i do the same in C++?

coor is already a Mx2 Mat, img is a completely different Mat, i have been triying for a week or more and can't find a solution

Ok, so basically are the coordinates of a set of contours resized as it follows:

    coor = cnts.reshape((len(cnts), 2))#cnts is the contours

in C++ i have the folloing for the contours

    vector<Point>cnts;//Retrieve the vector nested in the contours

and then due to lack of reshape as vector i move it to Mat

    Mat reshap = Mat (cnts);

    Mat coor = reshap.reshape(1);

Until this point it has the same size in C++ than in python, and the same data (i'm using a simple image test) after that i dont know how to generate the output from here

img[np.array(coor[:, 1]), np.array(coor[:, 0])]

On OpenCV C++

*EDIT: Think is working only modify to

Mat pixels; // initially empty
vector<Point> test;//Retrieve the vector nested in the contours
std::copy(pointList[ii].begin(),pointList[ii].end(),back_inserter(test));//Copy the vector the current n-th contour [ii]  and then your code 
    for (size_t i=0; i<test.size(); i++) {
  //  Vec3b p = image.at<Vec3b>(test); // bgr image
    Scalar p = image.at<uchar>(test[i]); // grayscale version//Modify here for gray scale
    pixels.push_back(p);
    }

then access to the required value as p.val[0];

Need to check with my other images but at this time think is working