Premium
An adaptive detection system for Raman spectral peaks at low signal‐to‐noise ratios
Author(s) -
Dyer Stephen A.,
Hardin David S.,
Bradley Eugene B.
Publication year - 1984
Publication title -
journal of raman spectroscopy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.748
H-Index - 110
eISSN - 1097-4555
pISSN - 0377-0486
DOI - 10.1002/jrs.1250150609
Subject(s) - raman spectroscopy , detector , estimator , additive white gaussian noise , noise (video) , gaussian , signal (programming language) , gaussian noise , a priori and a posteriori , raman scattering , white noise , signal to noise ratio (imaging) , physics , optics , statistics , algorithm , computer science , mathematics , artificial intelligence , quantum mechanics , philosophy , epistemology , image (mathematics) , programming language
Raman spectroscopic techniques have made it possible to study adsorbed species on single metal crystals at submonolayer coverages. However, as the amount of coverage decreases, the signal‐to‐noise ratio (SNR) deteriorates. This paper describes an adaptive peak detector (APD), a data analysis system capable of detecting peaks at low SNRs. The basic APD consists of anadaptive linear predictor (ALP) followed by a variance estimator. This detector is robust in that its performance does not rely on a priori knowledge of the exact signal and noise statistics. An experiment was performed in which single Lorentzian peaks embedded in white Gaussian noise at low SNRs were presented to the APD. A dramatic improvement in the detectability of the peaks is demonstrated; the experimental results indicate that the APD is useful at SNRs as low as 0.3. ALP convergence is shown to be a function of the input variance and the length of the predictor. Practical APD implementation considerations are given, and implications for future studies of adsorbed species are discussed.