Hi,

I'd like to use Eigen for solving the eigenvalue problem of a general matrix. I'll need this for a project using OpenCV (since the OpenCV solver only solves symmetric matrices), so I have to convert my data between both. Are there any best practices how to do so for dynamic sized matrices?

I don't have much experience with Eigen (so sorry if I am totally off here), but the first thing, that came to my mind is simply mapping the data of OpenCV's cv::Mat to a MatrixXd, like this:

Code: Select all

MatrixXd m = Eigen::Map<MatrixXd>(reinterpret_cast<double*>(a.data),a.rows,a.cols);


Which works fine for now, but I am afraid this is totally not best practice. Can anyone point me into the right direction or is it OK to do so?

Best regards, Philipp.

Sorry. My wrong. OpenCV already comes with a conversion defined in:

Code: Select all

#include <opencv2/core/eigen.hpp>


which makes it as easy as:

Code: Select all

cv::Mat_<float> a = Mat_<float>::ones(2,2);
Eigen::Matrix<float,Dynamic,Dynamic> b;
cv2eigen(a,b);


and is obviously a bit less error-prone than my attempt.

If someone is coming here from Google... I've pushed a simple example on how to do a Linear Discriminant Analysis with OpenCV and Eigen into my github repository at https://github.com/bytefish/opencv/tree/master/lda.