Localized sampling in the presence of noise (Q2488718)

The reconstruction of a band-limited function \(f(x)\) with bandwith \(\sigma>0\) i.e., a function in \(L^2(R)\) whose Fourier transform vanishes ouside the interval \([-\sigma,\sigma]\), making use of the Shannon sampling theorem, is a typical problem in data processing. In this note, the so-called ``localized sampling'' which is based upon the Shannon sampling theorem is used in order to approximate a band-limited function in the presence of noise. Error bounds are deduced with applications to various types of error when the sampled values are obtained via real-world acquisition devices.