Improving Absorbance Spectrum Reconstruction via Spectral Data Decomposition and Pseudo-Baseline Optimization

This paper addresses the problem of reconstruction of the optical absorption spectra on the basis of low-resolution data provided by a spectrophotometric transducer. It is shown that the performance of the algorithms of (generalized) deconvolution with a built-in positivity constraint, which is frequently used for this purpose, deteriorates for data containing a considerable baseline (background). A new method for overcoming that limitation is proposed. It consists of the subtraction of a rough estimate of the baseline (called pseudo-baseline) from the data, reconstruction of the spectrum on the basis of the data modified in this way, and restoration of the pseudo-baseline in the final result of spectrum reconstruction. The proposed amelioration is illustrated by means of an algorithm of spectrum reconstruction based on Jansson's method of deconvolution.