Title: Adaptive Savitzky-Golay filtering and its applications
Authors: József Dombi; Adrienn Dineva
Addresses: Institute of Informatics, University of Szeged, 6720 Szeged Árpád tér 2, Hungary ' Institute of Automation, Kandó Kálmán Faculty of Electrical Engineering, Óbuda University, 1034 Budapest Bécsi út 96/b, Hungary; Doctoral School of Computer Science, Department of Information Technologies, Universitá degli Studi di Milano, Crema Campus, 65 Via Bramante, Crema (CR), I-26013, Italy
Abstract: Noise reduction is a central issue of the theory and practice of signal processing. The Savitzky-Golay (SG) smoothing and differentiation filter is widely acknowledged as a simple and efficient method for denoising. However only few book on signal processing contain this method. As is well known, the performance of the classical SG-filter depends on the appropriate setting of the window length and the polynomial degree, which should match the scale of the signal since, in the case of signals with high rate of change, the performance of the filter may be limited. This paper presents a new adaptive strategy to smooth irregular signals based on the Savitzky-Golay algorithm. The proposed technique ensures high precision noise reduction by iterative multi-round smoothing and correction. In each round the parameters dynamically change due to the results of the previous smoothing. Our study provides additional support for data compression based on optimal resolution of the signal with linear approximation. Here, simulation results validate the applicability of the novel method.
Keywords: Savitzky-Golay filter; adaptive multi-round smoothing; iterative smoothing and correction; noise removal; data compression; digital smoothing polynomial filter; noise removal; data compression; linear approximation; intelligence paradigm; fuzzy set theory; EMG signal; fast-varying signal.
International Journal of Advanced Intelligence Paradigms, 2020 Vol.16 No.2, pp.145 - 156
Received: 29 Feb 2016
Accepted: 07 Nov 2016
Published online: 01 May 2020 *