Hyperspectral imaging: a review of best practice, performance and pitfalls for in-line and on-line applications

The objective of this paper is to provide a comprehensive review of best practice in hyperspectral imaging. The paper starts to review the taxonomy of the different spectral imaging techniques together with their advantages and disadvantages. The appropriate selection of cameras and spectrographs and their figures of merit are discussed and a detailed description is given of how to qualify and calibrate a pushbroom imaging system for on-line and in-line control. Special emphasis is given to detection and avoidance of specular reflection which can severely distort quantification of the spectral response. Recommendations for an ideal Lambertian illumination are given and the effects of scatter and absorption are discussed when particulate systems are investigated. Here, first principles are introduced and strategies for how to separate scatter from absorption are developed. A simple method using the Kubelka and Munk approach is examined and separated scatter and pure absorption spectra are shown. The same procedure is applied to show the lateral distribution of the separated scatter and absorption properties of an active pharmaceutical ingredient embedded in an excipient. The terms penetration and information depth are discussed and an example of penetration depth profile over wavelengths is provided. Based on a good quality optical set-up and a validated measurement procedure, a practical procedure is described to analyse the data cube using the chemometrics toolbox for hyperspectral imaging. Finally, a survey on selected applications demonstrates the future potential of hyperspectral imaging.