Chemometric tools for image analysis

Anna de Juan, Marcel Maeder, Thomas Hancewicz, Ludovic Duponchel, and 

Infrared and Raman Spectroscopic Imaging. Edited by Reiner Saizer and Heinz W. Siesier Copyright 2009 WiLEY-VCH Veriag GmbH Co. KGaA, Weinheim iSBN 978-3-527-31993-0 

Hyperspectral images represent a particular type of measurement that contains both spahal and spectral-and thus chemical-information about a sample. The sample is physically preserved and compartmented into small surface or volume areas, referred to as pixels or voxels , respectively. Each of these small portions of the sample is represented by a spectrum. Information on the chemical composition can then be extracted through an examination of the spectra and an interpretation of their bands, or by confrontation with a spectra library. Information on the constituent distribution is consequently obtained from the pixel-to-pixel spectral variation. 

It is important, therefore, to take into account the spectral and spatial information contained in an image, because this duality conforms the singularity of this type of measurement as opposed to other bulk sample spectroscopic measurements. Data analysis tools should also take this spatial/spectral character into account if the maximum information is to be extracted from the raw data. 

---

Basically, the book can be subdivided into three parts. In the first part the fundamentals of the instrumentation for infrared and Raman imaging and mapping and an overview on the chemometric tools for image analysis are covered in two introductory chapters. The second part comprises the chapters 3-9 and describes a wide variety of applications ranging from biomedical via food and agriculture to polymers and pharmaceuticals. Some historical insights are given as well. In the third part the chapters 10-15 cover special methodical developments and their utiHty in specific fields. 

I 2 Chemometric Tools for Image Analysis Beer-Lambert law. 

---