Simulate a Computed Tomography Imaging Spectrometer image from a 3D hyperspectral cube.
Input : A 3D image (preferably in the form of a .tiff file) representing a hyperspectral cube of an object. Output : A 2D image (.png) representing the CTIS acquisition of that object.
(image modified from Wikipedia)
MWE :
python main.py -I <input directory with hyperspectral cubes> -O <output directory for CTIS images>
to try it out with the example cube from this repo :
python main.py -I ressources/input_cubes_small -O .
Customization :
- Change CTIS geometries with
--grating,--min_wavelength,--max_wavelength.
- Change spectral sensitivity of the sensor by providing a csv defining it. An example is at ressources/spectral_sensitivity.csv.
- Chose whether you're providing a hyperspectral cube or the intermediary cube directly with
--field_stop - The CTIS action can be modelized by a matrix S between the reduced hyperspectral cube and the CTIS image. This matrix can be computed with
--compute_system_matrix. It can be useful for reconstruction approaches. Note however that the computation time can be quite long.
- The CTIS system is from : Computed-tomography imaging spectrometer: experimental calibration and reconstruction results, Descour et al, Applied Optics (1995) (pdf)
- The diffraction grating geometries we implemented come from : Maximizing the resolution of a CTIS instrument, Hagen et al, Imaging Spectrometry (2006) (pdf)