Normalization |
Normalization is a re-scaling of the spectral intensity that may be necessary to compensate for various experimental effects for better presentation, comparison, and quantitative or qualitative analysis of the data. Mnova ElViS offers two different algorithms of spectrum re-scaling: •Standard Normal Variate •Normalization Standard Normal Variate Standard normal variate (SNV) is a popular transformation that treats each spectrum as a vector (S) in accordance with the following equation: SNV is particularly useful when the sample volume (physical or virtual, as for on-line analysis) is not stable and may change from measurement to measurement. It is also applied to eliminate the so-called „scatter effect“ often observed in Vis/NIR spectra of solid and powder materials obtained in diffuse reflectance mode. SNV-correction is typically applied to spectral series prior to quantitative analysis. Normalization Simple Largest Peak normalization to 1 is a standard (pre)processing tool for UVIR spectral type. It is typically applied in small series to make spectra acquired at different conditions better suitable for comparison, peak interpretation and qualitative analysis. In databases all spectra are typically normalized to the unit intensity. You can also apply spectrum wise integral and vector length normalizations by selecting the appropriate option from the Processing template scroll down menu: Check the Full X Range box to apply the normalization to the entire dataset. If you need to apply the normalization to a selected region of interest, just select the range values from the edit boxes. For Raman datasets, you can also apply a spectrum set dependent probabilistic quotient normalizaton (PQN). The reference spectrum computed internally can be chosen to be the mean or median spectrum of those selected (from the stacked items table). For further information about it, check this paper: Anal. Chem. 2006, 78, 4281-4290. |