The Research Of Baseline Correction Based On Curve Fitting And Filtering In FTIR-ATR

With the rapid development of national economy, the demand of food and drug safety is higher and higher.This requests us to know more about food and drug ingredients.In order to guarantee the safety of the product that is directly related to people’s life.On one hand we must be able to accurate qualitative and quantitative detection of product that Has been rolled off the output line.On the other hand also need to be able to achieves real-time monitoring of production process in the process of production,to strengthen the control of production of each link, to find problems as early as possible, to minimize security risks. Infrared spectroscopy in this area has a very important and widely used. Infrared spectral data preprocessing for accurate calibration spectrum and quantitative analysis is very important, especially the calibration measurement conditions, environmental changes caused by the spectral baseline drift is one of the key in the infrared spectrum analysis.According to the features of baseline drift noise, This papers puts forward an improved asymmetric least squares method for infrared spectrum signal baseline drift correction. At the same time puts forward the comprehensive processing median filtering, wavelet transform algorithm. That first of all, median filter was used to extract rough base drift signal, then use wavelet decomposition reconstruction method to smooth base drift further correction. Median filtering to extract rough outline base drift, but also leads to a distortion of the "step", this is the value computation of truncation errors. Experiment results show that median filtering, wavelet transform comprehensive processing algorithm can suppress the truncation errors.In this paper, a method commonly used in the interpolation method based on the interpolation of the spectrum for baseline drift correction. For common defects interpolation methods, this paper presents an improved partial least squares method for baseline drift correction. The traditional method of fitting interpolation baseline effect is very dependent on the choice of interpolation points characteristics, the effect is very fitting instability. Improved Partial Least Squares estimation method using the histogram of the background values of the original estimated background spectra, estimated by iterative calculation baseline drift higher applicability and stability.The simulation results also show that the method baseline drift filter effect is very obvious.