Know more

See our research topics >>

See our projets >>

Sensors developped within the team COMiC

The team COMiC develops measurement systems based on multivariate indirect optical sensors, generally using spectrometry (UV-VIS-NIR) and/or hyperspectral imaging.

The common approach to extract useful information (such as absorption and/or scattering properties) on a studied medium follows two steps:

• The acquisition of an optical signal after its interaction with the medium
• The construction of a statistical model, by a computer-based learning, to establish a relationship between the signal and the quantity of interest.

This statistical model is then used to predict, as precisely as possible, the quantity of interest based on the optical signal acquired on a new sample.

General principle of a multivariate indirect optical measurement

Multi/hyperspectral measurements combined with calibration models

The team COMiC studies, combining multi/hyperspectral measurements and calibration models, have proven to give excellent results since more than two decades.

However, optical measurements and calibration models do not systematically reach required criteria in terms of quality and robustness, because of:

• Complexity and diversity of new studied media (ground, micro-algua, sewage sludge, genetically modified plants...)
• Measurements in uncontrolled conditions (outside, reactors, methanisors...).

 In order to outperform these limits, the team COMiC focuses on the following topics:

• Optical developments to improve signal quality,
• Chemometrics method developments to improve model robustness.