Hyperspectral Data Analysis: Algorithms and Pre-Processors for Industrial Sorting Applications

Jul 20, 2023
Facebook X LinkedIn Email
Login  Register
About This Webinar
Hyperspectral imaging is a beneficial machine vision technology for various industries. It allows engineers to differentiate between materials with similar physical or visual characteristics, which traditional imaging technology cannot do. Typical applications involve sorting, grading, and quality control, for example in the food and recycling industries.

One of the main benefits of hyperspectral imaging is its high spatial and spectral resolution which enables the detailed characterization of materials in a non-destructive manner. A hyperspectral camera produces images where a spectrum is measured for each pixel, creating a hyperspectral data cube that comprises spatial and spectral information. While hyperspectral imaging provides more information allowing more accurate analysis, dedicated algorithms that can handle multivariable data must be employed to process the data effectively.

This presentation focuses on two different types of algorithms, a non-supervised algorithm, which doesn't require any prior knowledge of samples, such as principal component analysis (PCA), and a supervised algorithm, which requires training of the dataset to teach a model, such as partial least squares discriminant analysis (PLS-DA). Marmion gives examples of quantitative and qualitative modeling of hyperspectral data with different pre-processors.

Marmion shows the effects of the pre-processors, highlighting how they are used to produce models that are more robust and resilient to the most demanding operative conditions. Finally, he also shares how the models based on hyperspectral data work online and gives practical examples of successful real-time applications.

*** This presentation premiered during the 2023 Vision Spectra Conference. For more information on Photonics Media conferences, visit

About the presenter

Mathieu MarmionMathieu Marmion, Ph.D., holds a doctorate in physical geography from the University of Oulu in Finland and a double Master of Science degree in electrical engineering from the Grenoble Institute of Technology (Grenoble INP) in France and the Norwegian University of Science and Technology (NTNU). He has been working at Specim Spectral Imaging for the past 12 years first as a technical sales engineer and a sales manager before becoming the lead applications specialist.
Imagingmachine visionVision Spectrahyperspectral imaging
We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.