I have an OpenCV program in C++ which takes a large Mat and returns a vector of smaller Mats, which I am trying to use in MatLab using mex (specifically mexOpenCV from here: https://github.com/kyamagu/mexopencv). I can get a single Mat back into the plhs[0] simply with plhs[0]=MxArray(theMats[0]) for example, but how can I return the whole vector?
Thank you!
#include "mexopencv.hpp"
void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
{
// Check number of arguments
nargchk(nlhs<=1 && nrhs==1);
// Convert mxArray to cv::Mat
cv::Mat mat = MxArray(prhs[0]).toMat();
std::vector<cv::Mat> theMats;
int ySize = 400;
int xSize = 400;
int yStride = ySize;
int xStride = xSize;
int cols = (mat.cols-xSize)/xStride + 1;
int rows = (mat.rows-ySize)/yStride + 1;
for (int i=0; i<cols; i++){
for (int j=0; j<rows; j++){
cv::Rect crop(cv::Point(i*xStride,j*yStride),
cv::Point(i*xStride+xSize, j*yStride+ySize));
theMats.push_back(mat(crop));
}
}
// Convert cv::Mat back to mxArray
plhs[0] = MxArray(theMats[0]); //I want theMats, not just theMats[0]
}
In the general case where cv::Mat may each be of different size, returning each cv::Mat as a matrix and returning the collection as a cell array is probably your only and best option:
https://www.mathworks.com/help/matlab/apiref/mxcreatecellarray.html
If each of the cv::Mat are of the same size, you could also entertain returning the collection of cv::Mat as an N-D numeric matrix where one of the dimensions is each cv::Mat and the remaining dimensions are the dimensionality of each of the cv::Mat objects.