CODEWITHSUNDEEP
c++opencvmatrix-multiplication
user1953728
user1953728

Reputation: 3

Opencv tensor product

I have two 2 x 2 matrices and I want to get the tensor product of them using opencv, for example:

{
{0, 1},
{1, 0}
}
x
{
{1, 0},
{0, 1}
}

=

{
{0, 0, 1, 0},
{0, 0, 0, 1},
{1, 0, 0, 0},
{0, 1, 0, 0}
}

So I have something like:

cv::Mat mat1 = (cv::Mat_<double>(2,2) << 1, 0, 0, 1);
cv::Mat mat2 = (cv::Mat_<double>(2,2) << 0, 1, 1, 0);

is there some tensor product function built into opencv to get the result I require?

Upvotes: 0

Views: 1382

Answers (1)

Miki
Miki

Reputation: 41775

OpenCV doesn't provide builtin operations on tensors. You need to rely on other libraries for that.

However, I'm no expert in tensors, but this looks like a Kronecker product to me (still no available in OpenCV). If so, you can find an implementation here that I report also in the code below, that produces the result you require:

[0, 1;
 1, 0]

 x

[1, 0;
 0, 1]

 =

[0, 0, 1, 0;
 0, 0, 0, 1;
 1, 0, 0, 0;
 0, 1, 0, 0]

Code:

#include <opencv2/opencv.hpp>
using namespace cv;
using namespace std;

Mat kron(const Mat& A, const Mat& B)
{
    CV_Assert(A.channels() == 1 && B.channels() == 1);

    Mat1d Ad, Bd;
    A.convertTo(Ad, CV_64F);
    B.convertTo(Bd, CV_64F);

    Mat1d Kd(Ad.rows * Bd.rows, Ad.cols * Bd.cols, 0.0);
    for (int ra = 0; ra < Ad.rows; ++ra)
    {
        for (int ca = 0; ca < Ad.cols; ++ca)
        {
            Kd(Range(ra*Bd.rows, (ra + 1)*Bd.rows), Range(ca*Bd.cols, (ca + 1)*Bd.cols)) = Bd.mul(Ad(ra, ca));
        }
    }
    Mat K;
    Kd.convertTo(K, A.type());
    return K;
}

int main(int argc, char **argv)
{
    Mat mat1 = (Mat_<double>(2, 2) << 0, 1, 1, 0);
    Mat mat2 = (Mat_<double>(2, 2) << 1, 0, 0, 1);

    Mat res = kron(mat1, mat2);

    cout << mat1 << endl << endl;
    cout << " x " << endl << endl;
    cout << mat2 << endl << endl;
    cout << " = " << endl << endl;
    cout << res << endl;

    return 0;
}

Upvotes: 1

Related Questions