Statistical Processing Methods for Polarimetric Imagery

Estimation theory is applied to a physical model of incoherent polarized light to address problems in polarimetric image registration, restoration, and analysis for electro-optical imaging systems. In the image registration case, the Cramer-Rao lower bound on unbiased joint estimates of the registration parameters and the underlying scene is derived, simpli ed using matrix methods, and used to explain the behavior of multi-channel linear polarimetric imagers. In the image restoration case, a polarimetric maximum likelihood blind deconvolution algorithm is derived and tested using laboratory and simulated imagery.Finally, a principal components analysis is derived for polarization imaging systems. This analysis expands upon existing research by including an allowance for partially polarized and unpolarized light.