Question

I am trying to convert the following c++ line into OpenCV matrix operation (which is also c++):

double myCode::calculate ( int i, int au )
{
    double k;
    for ( int j = 0; i < N; i ++ );
    {
         k += fabs(data[i][j] - means[au][j]);            
    }
}

I want to define “data” and “means” as openCV matrix type, like:

cv::Mat data ( NUMBER_OF_OBSERVATIONS, N, CV_8UC3 );
cv::Mat means = cv::Mat.zeros ( 5, N, CV_8UC3 );

then repeat the above class for this cvMat type “data” and “means”. How can I do that? Especially I don’t know how to do the line:

k += fabs(data[i][j] - means[au][j]); 

Thanks a lot.

Answer 1

You can simply write

double myCode::calculate ( int i, int au )
{
    cv::Scalar res = sum(avg(data(RowRange(i)) - means(RowRange(au))));
    return res[0] + res[1] + res[2]; // sum all the channels together
}

Note that RowRange() is not actually the correct syntax – look in OpenCV docs for the proper usage of Range(), but that’s the idea.