Computational and Hyperspectral Imaging
The group of computational and hyperspectral imaging in the TOPTEC Centre focuses on new imaging techniques based on so-called compressive imaging, as well as on the development of hyperspectral imaging systems in unconventional spectral ranges, such as the mid-infrared region.
Compressive sampling represents a new approach to data acquisition. It exploits the fact that many of the commonly measured datasets are sparse, i.e. they can be expressed by only a few significant components in a well-chosen basis. A typical example is a photograph, where a huge amount of information can be compressed via wavelet transformation into a small datafile (JPEG2000). Compressive sampling makes it possible to devise unconventional solutions to many problems, including imaging in “exotic” spectral areas.
Hyperspectral imaging denotes image acquisition where a spectrum is collected for each of the image pixels. We can, therefore, gain considerably more information about the scene compared to images acquired by a conventional camera. For instance, we can determine the concentration of particular chemical substances based on their spectral traces.
Team and equipment
- The group leader Karel Žídek has been engaged in time-resolved laser spectroscopy and implementation of compressive sampling in laser spectroscopy on a long-term basis. He was awarded the Otto Wichterle Award in 2017 for his work in ultrafast femtosecond spectroscopy.
- The group consists of several research workers (Ph.D. and B.Sc. degree) and Ph.D. students.
The group uses well-equipped laser laboratories, featuring:
- lasers (VIS to NIR) and incoherent light sources for VUV (160 nm) to far-IR
- detectors: UV-Vis-NIR spectrometers, UV-Vis-NIR photodiodes, photomultiplier modules, mid-IR camera, CMOS cameras
- specialized optical elements, e.g. light modulation via DMD (digital micromirror device)
The main areas of our research focus are:
- hyperdimensional imaging (combining imaging with spectral or temporal information), in particular for the exotic spectral (mid/far IR) and temporal (fs) ranges, where the application of standard approaches is problematic or even impossible. We aim at developing the use of uncooled detectors for hyperspectral imaging in the IR spectral region and we also search for new unconventional approaches to hyperspectral imaging by using so-called compressive sampling;
- computational imaging enabling us to develop novel imaging systems based on compressive sampling, where alternative approaches to image acquisition can be used (e.g. lensless imaging). We focus on (i) the CASSI technique (coded-aperture single-snapshot spectral imaging), which makes it possible to capture and reconstruct a 3D hyperspectral datacube by using a single 2D snapshot from a camera, and (ii) the single-pixel camera technique, which makes it possible to carry out imaging by using a single-pixel detector, such as a photodiode. We concentrate on possible implementations of computational imaging into laser spectroscopy.
- RODES (Robust Detection System with Hyperspectral Sensor) PDF (0.5 MB)
- Hyperspectral Detection System of Hazardous Substances PDF (0.5 MB)
- Coherent Excitation Coding for Exploitation of Compressed Sensing in Laser Spectroscopy PDF (0.3 MB)
- Imaging of chemical compounds with a hyperspectral camera
We collaborate with research institutions worldwide (Department of Chemical Physics, Lund University, Sweden), as well as within the Czech Republic (Thin Film Department, Institute of Physics of the CAS; Technical University Liberec; Military Research Institute, University of Defence in Brno). Our research has also benefitted from cooperation with such industry partners as APPLIC s.r.o. or CRYTUR, spol. s r. o.
RNDr. Karel Žídek, Ph.D.
Tel.: +420 487 953 901