Research On Denoising And Baseline Correction Algorithm Of Chromatogram

Chromatography is a powerful separation and detection method.In substance,it's a repeated process of distributing balance between stationary phases and mobile phases.In the process,different components in the sample can be separated from each other eventually,so that the sample can be analyzed qualitatively and quantificationally.Nowadays,chromatography which been widely used in the fields of food,medicine,chemical,material,environmental monitoring and so on provides convenience for human life.In the process of chromatographic separation,the influence of noise interference and baseline drift exist usually so that an ideal separation result is achieved difficultly.The reduction of the accuracy of the subsequent peak detection and peak resolution leads to a difficult actual sample detection.Noise interference comes mainly from the process of transforming analog signal into digital signal which can be dealt with conputer.A complex chromatogram after denoising should keep peak at original height basically and be smooth in placid area.However,single signal denosing method is difficult to accomplish concurrently.First,this paper uses the wavelet soft threshold method to denoise original chromatography,and then uses the variance of the local signal to distinguish between peak area and placid area.Finally,the Savitzky-Golay filter is used to denoise in placid area.The result of experiment shows that the signal-to-noise ratio is improved evidently.The chromatographic baseline is mainly derived from the objective environmental conditions in the process of seperation,such as the pollution in the mobile phase and the change of temperature and so on.In general,baseline drift appears at the bottom of the final chromatogram,and it will seriously affect the calculation of the height or area of the peak.It is found that the height of peak is relevant to the effect after baseline correction and the number of iterations in experiment.Therefore,the morphological method,local minimum method and the moving basepoint method are proposed to remove the peaks in this paper.Then the chromatogram after preprocessing will be dealt with adaptive iterative reweighted penalized least squares algorithm.In addition,this paper puts forward a new correction algorithm based on Savitzky-Golay in the condition of big window.They are more effective to reflect the true baseline trend than airPLS.The result of experiment proves that all the proposed methods can get a satisfactory effect in practical use.