Misa Hirose1, Rina Akaho2, Chikashi Maita1, Mai Sugawara1 and Norimichi Tsumura1
1 Graduate School Advanced Integration Science, Chiba University, CHIBA, JAPAN.
2 Department of Informatics and Image Sciences, Chiba University, CHIBA, JAPAN.
Optical Review, Volume 23, Issue 3, pp. 544–548 (2016).
Abstract
In this paper, spectral sensitivities of mosaic 5-band camera are designed
by using numerical skin phantom for separation of chromophore densities,
shading and surface reflection. To make the numerical skin phantom, spectral
reflectance of skin various was calculated by Monte Carlo simulation for
photon migration from the values of melanin, blood volume and oxygen saturation.
In the numerical skin phantom, melanin and hemoglobin concentration distribution
were obtained by separating from actual skin image by independent components
analysis and the calculated components were assigned to concentration distribution.
Utilizing this numerical skin phantom, spectral sensitivities of mosaic
5-band camera were optimized by using nonlinear optimization technique
to estimate spectral reflectance for skin separation. In this optimization,
the spectral sensitivities are assumed as normal distribution and sensor
arrangement is set as same as that of conventional mosaic 5 band camera.
From the results of optimization, we found that the optimized spectral
sensitivities can improve the spectral estimation significantly.
[PDF] (to appeare)