Optimized Spectroscopic Signal Estimates Using Integration and Matched Filters
This paper examines theories of signal processing as applied to peak magnitude estimation in absorption and emission spectroscopy. Signals obtained from Fourier transform, fixed wavelength, and scanning dispersive instruments are modeled in terms of the time required to obtain a spectrum. The differences between these techniques and the signal processing procedures that should be used for each technique are characterized for a Lorentzian spectral feature. Including the time required to scan over a range of optical frequencies results in optimal signal processing procedures that are different from those previously supposed. In particular, it is found that the optimal matched filter is less efficient than repetitive measurements at a single frequency. The theory developed for the Lorentzian line model is extended to include an arbitrary shaped peak.
Optimized Spectroscopic Signal Estimates Using Integration and Matched Filters Stephen E. Bialkowski Applied Spectroscopy 42 807 1988