search

Smorodov / LogPolarFFTTemplateMatcher

OpenCV Fourier-Mellin algorithm implementation.
MIT License
169 stars 45 forks source link

cooooool project #12

Closed tscmga closed 2 years ago

tscmga commented 2 years ago

so cool. sorry for my poor english maybe 1.normailize(f0,min_max) normailize the magtiue of dft , i think will make noise bigger

  1. add a hanning window to orign image border ,soft image bounder ,may good
  2. canny a image , can , change to use gradien of image , that is use sobel to get gx, gy...and to complex, dft..
tscmga commented 2 years ago

cv::normalize(f0, f0n, 0, 1, NORM_MINMAX); i think will make noise bigger

tscmga commented 2 years ago

and cache the log polar of one image is good ...if...use one template...cache the log polar dft of the template image is good

tscmga commented 2 years ago 
Mat planes[] = { Mat_<float>(padded), Mat::zeros(padded.size(), CV_32F) };
Mat complexImg;
merge(planes, 2, complexImg);
dft(complexImg, complexImg

set planes[0]=gx planes[1]=gy i think is the same to canny a image...but less parameters

Smorodov commented 2 years ago

Hi, thanks for your notes! Canny is more complex algorithm and it is not differentiating only, It has postprocessing. Really this project is just a base for experiments, post- and pre- processing is highly sensitive to the task you solving. Good luck !