Application of atomic functions in problems of signal preprocessing

It is known that information obtained from wide-spectrum measuring devices, especially in millimeter and sub-millimeter wavelength range, presents a set of discrete values. In practice, these values can be heavily distorted with noise. In this paper a problem of obtaining information about physical process with high probability, demanding culling of rough errors and smoothing of random errors, is investigated. For solving of such problems Kalman recursive algorithms of filtration, based on utilization of static signal parameters on measurer´s input, are used. After error smoothing procedure is applied, further signal processing with Kalman algorithms is performed. Statistical signal parameters after rough error removal and random error smoothing are determined with classic methods.