Proceedings Article | 7 March 2016
KEYWORDS: Image filtering, Birefringence, Optical filters, Bandpass filters, Linear filtering, Polarizers, RGB color model, Biology, Materials science, Environmental science, Transmittance, Liquid crystals, Microscopes, Polarization
While a pixel in a color image has three colorimetric information of RGB, that in a spectral image contains full spectral
information, several tens times more information compared to the color image. Hence, the spectral image is widely
applicable in biology, material science, and environmental science. Although several methods for spectral image
acquisition have been suggested to date, those methods are expensive, bulky, or slow in actual device. In this work, we
designed a novel type of tunable narrow band-pass filter using rotatable polarizer, quarter-wave plate, and birefringence
films. Different from the conventional Lyot-Ohman type filter, we do not use a liquid crystal layer. The selection of
wavelength is made by rotating the polarizer in our filter set, and adopted a piezoelectric rotational actuator for that. We
simulated to find the optimal conditions of the filter set, and finally, fabricated a filter module. The minimum band width
was 5 nm, which is suitable for usual spectral imaging and can be reduced further if necessary, and the wavelength of
light passing through the filter set was continuously selectable. After setting the filter in a microscope, we obtained a
spectral image set for a bio sample that contained full spectrum information in each pixel. Using image processing, we
could demonstrate to read out the spectral information for any selected position.
