Hyperspectral Image Sensor Development Using Compressive Sensing on Multispectral Pixel Arrays

Hyperspectral imaging delivers detailed spectral information far beyond traditional RGB or multispectral sensors, enabling powerful applications in precision agriculture, biomedical diagnostics, remote sensing, and industrial inspection. However, conventional hyperspectral systems require bulky optics, large data bandwidth, and complex acquisition schemes, limiting their deployment in compact and low-power environments.

As we move into the Edge AI era, sensors are increasingly expected to perform efficiently in real-time, on-device systems such as autonomous drones, mobile platforms, and smart medical instruments. This creates a strong need for high-resolution spectral sensing with minimal data output.
In this internship, we propose an innovative solution by combining compressive sensing techniques with a multispectral pixel array. This allows for accurate reconstruction of hyperspectral data using a reduced number of readout channels, significantly lowering the sensor’s output data volume and energy consumption. At the same time, by retaining a multispectral filter array approach, the system benefits from CMOS compatibility, scalable integration, and simple optical design, offering a practical path to miniaturized hyperspectral imagers.

The student will contribute to the design and simulation of pixel architectures, explore spectral encoding strategies, and implement reconstruction algorithms based on sparsity or deep learning. The work will include device simulation, circuit design and algorithm validation using IMEC’s internal tools and test platforms.
This internship provides hands-on experience at the intersection of hardware design and computational imaging. Background knowledge in image sensors, optics, signal processing, or machine learning will be considered a strong asset.

Type of Project:  Combination of internship and thesis

Master's degree:  Master of Engineering Technology

Master program: Electrotechnics/Electrical Engineering

Duration: 1 Year

Supervisor: Jan Genoe (EE, Nano)

For more information or application, please contact the supervising scientist Myonglae Chu (myonglae.chu@imec.be).

Only for self-supporting students.