Identification of Different Pretreatment Methods of Rice Raman Spectroscopy
author: Christ
2022-01-06
Identification of Different Pretreatment Methods of Rice Raman Spectroscopy
It is difficult for consumers to identify the regional brand rice instead of unique geographical factors.Based on Raman spectroscopy technology, test comparison different pretreatment methods include first derivative + translation smoothing, second derivative + translation smoothing, wavelet transform + remove baseline three commonly used pretreatment methods, also proposed an improved segmented polynomial fitting + remove baseline a total of four pretreatment methods, combined with partial least squares method to achieve identification analysis of rice, put forward a best pretreatment method of rice.First, 150 samples of rice ranging from 200 to 3100 c m-1 in Yi'an County, Heilongjiang Province were collected by Raman spectrometer, and then raw Raman profiles were preprocessed with first derivative + translation smoothing, second derivative + translation smoothing, wavelet transform + baseline removal, segmented polynomial fitting + baseline removal.We selected 33 samples from each locarea for training, and established a differential analysis model based on partial least squares for unknown 51 samples. In the training set, the largest, the least, and root mean square error, and the worst in the test set.Finally, through the PLS modeling results, In, Training, practice, set, in, The total identification rate of rice in all three kinds of rice was 100%; In the test set, The total identification of rice in 100% using 3:2 fitting + removal baseline was 100%, The total identification rate of rice from three origin using 5 points 2 fitting + removal baseline was 52.9%, Other piecewise polynomials fit in between; The total recognition rate of using the first derivative + translation smoothing, second derivative + translation smoothing and wavelet transformation was 88.2%, respectively, 86.2% and 96.1%; It is found, The advantage of 3 point 2 fits + in segmented polynomial fit is obvious, Consiagreement with its correlation coefficient, mean square error and root mean square error, High overall identification rate, The discrimination effect was stable.
Principle
The Raman spectrum identifies the molecular structure by varying the molecular scattering intensity of visible monochromatic light inside the material, thus a specific fingerprint calibration of the internal functional group of the matter, and the spectral peak intensity is related to the molecular concentration.Raman spectroscopy is used for the detection of agricultural products, mainly focusing on the research of food, dairy products, fruits and vegetables, edible oil, etc.
Raman spectroscopy solution:
Spectrum was collected using a ATR-3100-785nm Raman spectrometer manufactured by OPTOSKY(Xiamen) with a detection range of 124.79 ~ 3324.66cm-1. Under the optimal measurement conditions, the displacement value deviation of the measurement standard peak was zero, and the displacement accuracy does not exceed ± 4cm-1.
Raman spectroscopic detection parameters were set as: laser power 300mW, excitation wavelength 785nm,resolution 6.58cm-1, integration time 5000ms, scanning range of 200 to 3300 c m-1.
Principle
The Raman spectrum identifies the molecular structure by varying the molecular scattering intensity of visible monochromatic light inside the material, thus a specific fingerprint calibration of the internal functional group of the matter, and the spectral peak intensity is related to the molecular concentration.Raman spectroscopy is used for the detection of agricultural products, mainly focusing on the research of food, dairy products, fruits and vegetables, edible oil, etc.
Raman spectroscopy solution:
Spectrum was collected using a ATR-3100-785nm Raman spectrometer manufactured by OPTOSKY(Xiamen) with a detection range of 124.79 ~ 3324.66cm-1. Under the optimal measurement conditions, the displacement value deviation of the measurement standard peak was zero, and the displacement accuracy does not exceed ± 4cm-1.
Raman spectroscopic detection parameters were set as: laser power 300mW, excitation wavelength 785nm,resolution 6.58cm-1, integration time 5000ms, scanning range of 200 to 3300 c m-1.
Results
Fig 1.Raw Raman spectrum of three produing area of rice
Fig 2. Main characteristic peaks of rice Raman spectrum
Fig 3. Pre-processing method of first derivative+translation smoothing
Fig 4. Pre-processing method of second derivative+translation smoothing
Fig 5. Pre-processing method of wavelet transform+baseline removal
Fig 6. Pre-precessing method of piecewise polynomial fitting+baseline removal
Conclusions:
Raman spectroscopy combined with different pretreatment methods to identify three similar origin of rice, respectively, using first derivative + shift + translation smooth smooth, the second derivative, wavelet transform + methods for spectra pretreatment to remove the baseline, because these methods cannot keep the original wave shape or the phenomenon of baseline drift, A pre-processing method of piecewise polynomial fitting + removing baseline was proposed. The partial least squares PLS method was used to establish the Raman model of 150 samples of rice from three producing areas. Experimental results showed that the model established after piecewise polynomial fitting + removing baseline of 3 points in the pre-processing of quadratic polynomial was the most accurate. In the training set and test set, the recognition rate is 100%, and the clustering effect is good. The pretreatment of 3-point quadratic polynomial + removing baseline provided an effective method for identification of rice from similar areas, and also provided technical reference for identification of other crops in nearby areas.
Raman spectroscopy combined with different pretreatment methods to identify three similar origin of rice, respectively, using first derivative + shift + translation smooth smooth, the second derivative, wavelet transform + methods for spectra pretreatment to remove the baseline, because these methods cannot keep the original wave shape or the phenomenon of baseline drift, A pre-processing method of piecewise polynomial fitting + removing baseline was proposed. The partial least squares PLS method was used to establish the Raman model of 150 samples of rice from three producing areas. Experimental results showed that the model established after piecewise polynomial fitting + removing baseline of 3 points in the pre-processing of quadratic polynomial was the most accurate. In the training set and test set, the recognition rate is 100%, and the clustering effect is good. The pretreatment of 3-point quadratic polynomial + removing baseline provided an effective method for identification of rice from similar areas, and also provided technical reference for identification of other crops in nearby areas.
Microscopic Raman Spectroscopy for Graphene Sample Testing - Raman Spectroscopy and its Most Common Applications
Raman : Applications 
Application of Raman Spectroscopy in Characterization of Graphene Structure 
Food Analysis and Raman Spectroscopy 
Raman or FTIR Spectroscopy 
What is Raman Spectroscopy? 
Raman spectroscopic detection and analysis of Scutellaria baicalensis herbal tablets