Spectral reflectance estimation of an object via low-dimensional snapshot requires both image acquisition and a post numerical estimation analysis. In this study, we set up a system incorporating a homemade cluster of LEDs with spectral modulation for scene illumination, and a multi-channel CCD to acquire multichannel images by means of fully digital process. Principal component analysis (PCA) and pseudo inverse transformation were used to reconstruct the spectral reflectance in a constrained training set, such as Munsell and Macbeth Color Checker. The average reflectance spectral RMS error from 34 patches of a standard color checker were 0.234. The purpose is to investigate the use of system in conjunction with the imaging analysis for industry or medical inspection in a fast and acceptable accuracy, where the approach was preliminary validated.

Date Published: 23 September 2013
PDF: 9 pages
Proc. SPIE 8870, Imaging Spectrometry XVIII, 88700C (23 September 2013); doi: 10.1117/12.2023263

Published in SPIE Proceedings Vol. 8870:
Imaging Spectrometry XVIII
Pantazis Mouroulis; Thomas S. Pagano, Editor(s)