Some Observations on the Resolution Enhancement of Spectral Data by the Method of Self-Deconvolution

The directly observable information in a digitally encoded spectrum can be selectively enhanced by deconvolution with an appropriate filter operator. This may be carried out on the frequency (wavenumber) domain spectrum or on its Fourier transform. Depending on the choice of the filter, one may achieve: (1) reduction in the random noise, (2) reduction of systematic instrumental errors, (3) enhancement of the resolution. These general deconvolution procedures are briefly reviewed in terms of this classification. Recent work at the National Research Council of Canada at Ottawa has emphasized the application to band resolution enhancement (self-deconvolution), principally with respect to Fourier transform infrared spectra. Experience in our laboratory with infrared data generated on dispersion spectrometers is described. Our results confirm those reported by Kauppinen et al. at Ottawa, but the need for caution in applying self-deconvolution in routine laboratory measurements is stressed.