Quantitative Analysis Of Spectral Data By Combined Deconvolution And Curve-Fitting

Curve-fitting (CF) and deconvolution are both established techniques for obtaining quantitative information from complex spectra, but they both have limitations. If CF is to be applied to heavily overlapped bands it is often impossible to determine the number of component bands present, and the band parameters obtained can be subject to large errors. The main disadvantage of deconvolution is that the choice of the parameters is often very subjective, and if incorrect can lead to erroneous results. In this work the advantages of combining the two techniques are investigated, and it is demonstrated that the problems outline above can be largely overcome. A comparison is made in this context of the relative merits of two methods of deconvolution: Fourier self deconvolution and maximum likelyhood restoration.