Smoothing of Data by Least Squares Procedures and by Filtering
Marcel J. E. Golay · IEEE Transactions on Computers · 1972
It is shown that when discrete experimental data are smoothed by fitting 2m + 1 consecutive data to a polynomial of 2nth degree, with n≪m, and when n and m are increased indefinitely, the smoothing obtained is equivalent to passing the original data through an ideal low-pass filter.