Spectrum segmentation and spectrum comparison method based on spectrum peak-splitting fitting

Abstract

The invention discloses a spectrum segmentation and spectrum comparison method based on spectrum peak-splitting fitting, and the method comprises the steps: carrying out the fitting of a spectrum peakin a spectrum of an unknown object through a mathematic function, and carrying out the segmentation of a fitted spectrum peak parameter, and obtaining each spectrum segment and the start and end positions of the corresponding wavelength; performing the same segmentation on a standard spectrum according to the starting position and the ending position of the spectrum segment to obtain the spectrumsegment of the standard spectrum; fitting the spectrum segments of each pair of unknown object and standard object with the related wavelength range, and correcting the spectrum of the unknown objectby using fitting parameters; and respectively combining the corrected unknown object spectrum segment and the standard spectrum segment, reconstructing the whole spectrum, and calculating the similarity between the reconstructed unknown object spectrum and the standard spectrum. The method can be applied to comparison of an infrared spectrum, a near infrared spectrum, a Raman spectrum and an ultraviolet-visible spectrum, and the positive judgment rate of spectrum comparison can be remarkably improved.

Description

technical field

[0001] The invention belongs to the field of spectral data processing, in particular to a spectrum peak segmentation and spectrum comparison algorithm. Background technique

[0002] As important molecular spectroscopy techniques, Raman spectroscopy and infrared spectroscopy have been widely used in structural identification and rapid identification of substances, such as the rapid identification of drugs, explosives, and chemical raw materials. The usual method of spectral identification is: establish a spectral database (generally pure compounds); collect spectra for unknown substances, compare them with the spectra in the database, calculate the correlation coefficient, and judge the unknown and the standard according to the size of the correlation coefficient (database). material) matching degree. Calculation of correlation coefficient is one of the most important links in substance identification, and it is also called similarity degree, matching degree,...

Images