Skip to Main Content
A new methodology for optimized and automatic spectral calibration is proposed. It is aimed to work in any spectroscopic system (linear or imaging) and comes to solve uncertainty problems in the quantification of transmittance and absorbance parameters when the detection sensor offers a poor signal to noise ratio. The goal is to automatically crop the spectral range removing those wavelengths where the application of the conventional spectral correction produces errors from a chemical point of view. Two cropping alternatives are presented and compared: one based on a factor related with the dark noise present in the black reference used for the spectral correction, and another related with the allowed variance in the measurement of the white reference of the spectral correction. The proposed procedures will enhance the performance of automatic wavelength and feature selection algorithms.