I've recently discovered this article : Article Fourier ring correlation as a resolution criterion for super...
about FRC and its use for resolution cutoff criterion in microscopy. In this paper they cut their localisation data in two halves and then use them to construct an image on which they use Fourier Transformation to calculate Fourier Ring Correlation.I wonder why one does not directly calculates it from the set of datas [x y] which could be used for 2D discrete fourier transform. This would also have the advantage of not depending of the pixel size of the image. (meaning: if two localisation are close to each other and should occupy the same pixel in the final image, do you sum them ? if you don't you lose information but if you do you also add false information to your image... )
This bugg me as FRC on two reconstructed images seems less consistent than direct FRC on datas.
thanks for your help,