Pre-processing of NIR Åsmund Rinnan
Introduction NIR Fructose Glucose 25 50 75 75 50 25
Introduction NIR Baseline with a slope/ curve Nonlinearity Fructose Glucose 25 50 75 75 50 25 A “fat” baseline
Introduction Techniques Reference dependent O-PLS OSC OS SIS Reference independent MSC/ ISC EMSC/ EISC SNV Detrend Normalization Savitzky-Golay Norris-Williams Finite difference
Introduction Techniques Correction of light scatter MSC EMSC Detrend SNV Derivation Finite difference Savitzky-Golay
Effect of pre-processing Before Specular effect
Effect of pre-processing After
MSC Raw Reference
MSC Raw Reference
MSC Raw spectrum b = Slope a = Intercept Reference P Geladi, D MacDougal, H Martens (1985): Linearization and scatter correction for near-infrared reflectance spectra of meat, Applied Spectroscopy, 39, 491-500
MSC
Extended MSC Wavelength Slope & Intercept Known spectra correction = = Basic MSC = Detrend H. Martens, E. Stark (1991): Extended multiplicative signal correction and spectral interference subtraction: new preprocessing methods for near infrared spectroscopy, Journal of Pharmaceutical and Biomedicinal Analysis, 9, 625-635
Extended MSC The correction Calculated Step 1: Step 2: H. Martens, E. Stark (1991): Extended multiplicative signal correction and spectral interference subtraction: new preprocessing methods for near infrared spectroscopy, Journal of Pharmaceutical and Biomedicinal Analysis, 9, 625-635
X X Extended MSC Summary Parameter setting Paraemter setting MSC Detrend
SNV R.J. Barnes, M.S. Dhanoa, S.J. Lister (1989): Standard Normal Variate Transformation and De-trending of Near-Infrared Diffuse Reflectance Spectra, Applied Spectroscopy, 43, 772-777
SNV vs MSC MSC SNV
SNV vs. MSC Classificafication of barley
SNV vs. MSC SNV MSC Normalization Reference needed No reference Outlier sensitive Only similar spectra No normalization
Derivation
Derivation Finite difference
Savitzky-Golay vs. Finite difference Derivative Method 1 2 3 4 1st SG 0.38 (5) 0.40 (5) 0.47 (6) 0.46 (3) Finite difference 0.34 (5) 0.58 (4) 0.73 (4) 2nd 0.34 (6) 0.51 (4) 0.67 (5) 0.72 (4) 0.88 (5) 0.91 (5) 0.42 (6)
Derivation Savitzky-Golay A Savitsky, M J E Golay (1964): Smoothing and differentiation of data by simplified least squares procedures, Analytical Chemistry, 36 (8), 1627-1639
Derivation Savitzky-Golay Smoothing 11 9 7 5 3 A Savitsky, M J E Golay (1964): Smoothing and differentiation of data by simplified least squares procedures, Analytical Chemistry, 36 (8), 1627-1639
Derivation Savitzky-Golay Smoothing Polynomial 7 4 2 5 1 3 A Savitsky, M J E Golay (1964): Smoothing and differentiation of data by simplified least squares procedures, Analytical Chemistry, 36 (8), 1627-1639
Derivation Savitzky-Golay 7 point smoothing 15 point smoothing 2nd order derivative
Summary Pre-processing of NIR Run the following for near to optimal results MSC with Only 1st order reference correction or SNV 1st order reference correction and 2nd order wavelength correction Only 2nd order wavelength correction Savitzky-Golay derivation with 2nd order polynomial smoothing 7 points smoothing for the 1st derivative 9 points smoothing for the 2nd derivative
Compression factor Barley data Pre-processing MB MH Raw 0.0 % MSC1 88.0 % 90.4 % MSC2 92.3 % 94.4 % SNV SNV + Detrend 90.9 % 93.9 % SG1 91.6 % 92.8 % SG2 93.3 %