Summary: Z .Chemometrics and Intelligent Laboratory Systems 41 1998 3142
A structural approach for smoothing noisy peak-shaped analytical
A. Antonacopoulos a,)
, A. Economou b,1
Department of Computer Science, UniÕersity of LiÕerpool, Peach Street, LiÕerpool L69 7ZF, UK
Department of Instrumentation and Analytical Science, UMIST, PO Box 88, Manchester M60 1QD, UK
Received 16 March 1997; accepted 25 March 1998
This work describes an approach for smoothing noisy peak-shaped analytical data based on the identification of the struc-
tural form of the signal. The data series was described first as a succession of `peak structures' and then as a succession of
`meta-peak structures'. This description enabled convenient identification of the characteristic peaks arising from an analyti-
cal measurement and their separation from the noise components in the data. The method was applied to both voltammetric
and spectroscopic data featuring different distributions of noise in the frequency domain. It was demonstrated that the sug-
gested structural approach is successful in identifying the characteristic peaks with precision and, subsequently, in smoothing
the test signals. The smoothing operation relying on the structural approach is fast, and, in contrast to traditional smoothing
techniques, the fine detail of the signals is retained and no artefacts are generated as a result of the smoothing operation.
q 1998 Elsevier Science B.V. All rights reserved.