/Advancing outdoor hyperspectral imaging under challenging illumination conditions

Advancing outdoor hyperspectral imaging under challenging illumination conditions

PhD - Leuven | Just now

Unlock the true hyperspectral imaging capabilities in outdoor conditions.

Imaging technologies are rapidly becoming indispensable tools for measuring and monitoring in diverse outdoor applications – ranging from precision agriculture (tracking variability within and between crops) to infrastructure inspection (detecting corrosion or structural defects), and automotive systems (identifying pedestrians and objects on the road). Among these, hyperspectral imaging has proven to be far more powerful than conventional RGB imaging for many agricultural, industrial, and automotive use cases. 

 

The major challenge with outdoor imaging data is to interpret the measured data correctly in relation to the physical environment and to other captured data such as continuous point measurement devices. As opposed to controlled lab environments, the obtained imaging and hyperspectral data is susceptible to the varying incident solar radiation (even under cloudless skies), sun-sensor viewing angles (depending on time of day, season, and camera position), and scene geometry (flat surfaces versus complex structures). To complicate matters further, varying light conditions require sensors with very high dynamic range to avoid saturation. This can be addressed through strategies such as interleaved multi-exposure acquisition, auto-exposure techniques, or adaptive sensing solutions. All of these factors profoundly impact how data is analyzed and interpreted. 

 

The purpose of this PhD is to design a methodology to obtain highly trustworthy and interpretable imaging and hyperspectral data outdoor regardless of the varying light conditions. This encompasses the design of sensor fusion strategies with calibrated spectrometers measuring the incoming radiation as well as the design of novel data representations (extensions to BRDF models) and correction algorithms for radiances across different exposures, including the exploration of machine learning approaches that leverage reference objects with known spectra (e.g., road markers in automative scenes) to compensate for light variability across angles and scenes. This research will not only have a large impact on outdoor hyperspectral data quality, but also on the research in the different imaging technology under varying light condition (RGB imaging) and wide range of application domains (media and broadcasting, agriculture, health monitoring of architectural heritage, corrosion detection on high voltage towers, automotive). Finally, the student will have opportunities to learn the newest IMEC hyperspectral sensor technology, get hands-on experience with various of advanced imaging technologies, and grow into an expert in the field of image and signal processing outdoor. 



Required background: Computer Science / Physics of light and Optics / Data Science or equivalent

Type of work: 70% machine learning and modeling, 20% experimental, 10% literature

Supervisor: Hiep Luong

Co-supervisor: Wilfried Philips

Daily advisor: Mina Zahiri, Steven Thijs

The reference code for this position is 2026-028. Mention this reference code on your application form.

Who we are
Accept analytics-cookies to view this content.
imec's cleanroom
Accept analytics-cookies to view this content.

Send this job to your email