Spectral Image Acquisition with Rewritable Color Filters

We present techniques for spectral image acquisition. The spectral domain of the images is represented by a lowdimensional component image set, which is used to obtain an efficient compression of the high-dimensional spectral data. First, computational techniques to design color filters with a constraint of positive spectral values is described. Then, we present two prototypes of our spectral imaging systems that can be used to acquire the low-dimensional component image set optically through rewritable color filters. The first prototype is based on a spectral synthesizer that illuminates the sample with the light corresponding to a wanted color filter. The second prototype is based on a linear variable filter (LVF) and a liquid crystal spatial light modulator (LCSLM) that implements the rewritable color filter in front of a CCD-camera. We also show how liquid crystal tunable filter (LCTF) based system can be used to implement arbitrary color filters. The optically acquired component image set can be used for computational spectral image reconstruction or it can be directly used for pattern recognition tasks.