Open AccessColor reproduction from low-SNR multispectral images using spatio-spectral Wiener estimation
Author(s) -
Yuri Murakami,
Ken Fukura,
Masahiro Yamaguchi,
Nagaaki Ohyama
Publication year - 2008
Publication title -
optics express
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.394
H-Index - 271
ISSN - 1094-4087
DOI - 10.1364/oe.16.004106
Subject(s) - multispectral image , computer science , artificial intelligence , computer vision , spectral imaging , noise (video) , multispectral pattern recognition , color difference , spectral density estimation , signal to noise ratio (imaging) , channel (broadcasting) , optics , mathematics , image (mathematics) , physics , fourier transform , telecommunications , enhanced data rates for gsm evolution , mathematical analysis
It is possible to capture images with high-fidelity color by using a spectrum-based color image reproduction method that estimates the spectral information of objects from a multispectral image along with information from illumination. When multispectral images do not have a sufficient signal-to-noise ratio (SNR), the accuracy of the spectral and color estimation will be reduced. To improve color estimation accuracy, this paper proposes a spatio-spectral Wiener estimation, which uses spatial correlation as well as spectral correlation. In order to show the effectiveness of the proposed method, computer simulations and an experiment are carried out using a six-channel video camera. As a result, it is assured that the proposed method improves color estimation accuracy and suppresses color noise.