Pittcon Conference 2008

Middleton Research is presenting "Hyperspectral Identification of Fast Moving Materials with a New High-Speed Chemometric Method"

Abstract: Spatial measurement of concentration distribution or identification of components in stationary samples, such as tablets, is being done extensively in the pharmaceutical industry using 2D hyperspectral cameras. Continuously moving industrial samples, on the other hand, are much better suited to hyperspectral line-cameras producing full high resolution spectra of each point of a line at very high speed.

Current spectral software packages use distance based algorithms, such as spectral angle mapper, K-nearest neighbors, Mahalanobis distance and others to classify each spatial point to a class for identification purposes. Most of these algorithms require complex full spectrum calculations, which need to be performed at up to 34 kHz rate.

A new approach is proposed for identification of materials based on techniques and algorithms used for quantitative analysis. The results for this method will be presented with applications for identification of high value materials in a stream mix, and for web-based industrial hyperspectral quality assurance.

Request Presentation

Back to What's New