# fftshift implementation for OpenCV

I'm trying to implement fftshift from matlab for OpenCV. Can you please review the correctness of my algorithm? Have I missed something? Also, is there a better and faster way to do it?

cv::Mat ff;
cv::dft(distanceF, ff, cv::DFT_COMPLEX_OUTPUT);

//Make place for both the complex and the real values
cv::Mat planes[] = {cv::Mat::zeros(distanceF.size(),1, CV_32F), cv::Mat::zeros(distanceF.size(),1, CV_32F)};
cv::split(ff, planes);    // planes[0] = Re(DFT(I), planes[1] = Im(DFT(I))

cv::vector<float> im(planes[1].cols);
cv::vector<float> re(planes[0].cols);
int m = planes[0].cols;
int p = ceil(m/2);

for (int i = 0; i < p; i++)
{
im[i] = planes[1].at<float>(p + i +1);
im[p +i +1] = planes[1].at<float>(i +1);

re[i] = planes[0].at<float>(p + i +1);
re[p +i +1] = planes[0].at<float>(i +1);
}
im[p] = planes[1].at<float>(0);
re[p] = planes[0].at<float>(0);
cv::Mat input[] = {cv::Mat(im),cv::Mat(re)};
cv::merge(input,2,ff);

• You should replace the for loop with 2 copyTo() calls with the proper ROI set on the source and destination (use the () operator overload). Jan 5, 2016 at 14:33

I think the best way to implement fftshift is like it is done in OpenCv itself:

//The body is inspired from OpenCV, the fct_fftshift is mine
void fct_fftshift(cv::Mat& src)
{
int cx = src.cols/2;
int cy = src.rows/2;

cv::Mat q0(src, cv::Rect(0, 0, cx, cy));
cv::Mat q1(src, cv::Rect(cx, 0, cx, cy));
cv::Mat q2(src, cv::Rect(0, cy, cx, cy));
cv::Mat q3(src, cv::Rect(cx, cy, cx, cy));

cv::Mat tmp;
q0.copyTo(tmp);
q3.copyTo(q0);
tmp.copyTo(q3);

q1.copyTo(tmp);
q2.copyTo(q1);
tmp.copyTo(q2);
}

int main(int argc, char* argv[])
{