Vynogradov V. Wavelet analysis of multiscale processes and signals.

Based on a wavelet representation of 1/f-processes a method of detection of a 1/f-component on a white noise background is developed. A technique is proposed for finding parameters of individual pulses from time series, which are a superposition of a large number of overlapping pulses. Wavelet analysis of chaotic signals has revealed the existence of characteristic features, which can be used for identification and classification of such signals. A new algorithm is proposed and introduced of wavelet filtration of images, which is based on the application of template wavelet coefficients and multiscale analysis, for noise elimination in the vicinity of boundaries and enhancement of image contrast.