Abstract
A hyperspectral camera developed by Plain Sight systems allows the selection of spatial and spectral features. This camera uses a DMD to implement a non-conventional imaging process whereby the sensed image is a mathematical transform of the input scene. The Lockheed Martin team (including Yale, Duke and Johns Hopkins University) is currently working with this camera to jointly optimize and exploit the spatial and spectral degrees of freedom to detect targets in a scene. In the figure below, a horizontal slice through a typical raw data cube reveals that the DMD imaging systems has sensed a 1D Hadamard transform of each row of the input scene. The reconstructed data on the right is obtained by applying the inverse Hadamard transform to each row of the slice to recover the original 2D scene from the raw data. On the ISP program, we are exploiting this ability of the DMD to encode multi-scale wavelet-type transform for task specific operations.
© 2005 Optical Society of America
PDF ArticleMore Like This
Abhijit Mahalanobis and Robert Muise
FWN1 Frontiers in Optics (FiO) 2006
Nopparit Intharasombat, Tawei Ho, and Alexander A. Sawchuk
MP6 International Symposium on Optical Memory and Optical Data Storage (ODS) 2005
M. Alouini, A. Grisard, E. Lallier, D. Dolfi, F. Goudail, N. Roux, and Ph. Réfrégier
CTuA2 Computational Optical Sensing and Imaging (COSI) 2005