The color is not always sufficient to accurately represent the spectral reflectance properties of objects in natural scenes. For this reason, these scenes should be acquired by multispectral cameras. Because these systems are expensive and are sensitive to a limited number of bands, single-sensor snapshot systems have been recently developed. Most of them use a multispectral filter array (MSFA) laid over the sensor that spectrally samples the incident light, like the widely used Bayer color filter array (CFA) in color imaging. The MSFA is defined by a basic periodic pattern in which each filter is sensitive to a narrow spectral band. Each pixel of the resulting raw image is then characterized by one single band according to the MSFA. We presently work on:
- The spectral characterization of MSFA cameras used in uncontrolled conditions;
- The low-level MSFA image analysis (denoising, demosaicing) to obtain spectral cubes;
- The design of new spectral texture features.
⚈ Problem: 15 levels/pixel to be estimated for MSFA demosaicing instead of 2 levels/pixel for CFA.
What are our contributions?
MSFA image demosaicing by pseudo-panchromatic image: http://dx.doi.org/10.1109/TCI.2017.2691553
⚈ Problem: Spatial and spectral correlations within MSFA image are different from those in CFA images.
What are our contributions?
New multi-spectral texture database: https://doi.org/10.3390/s18072045
MSFA LBP for texture classification: https://dx.doi.org/10.1364/JOSAA.35.001532