A comparative assessment of various super-resolution techniques in target detection and enhancement using hyperspectral data

摘要

Algorithms for target detection in hyperspectral data successfully detect full pixel targets; but, they fail to simultaneously detect part of target which may be lying partially in surrounding pixels. This requires development of algorithms which can simultaneously detect the sub pixel targets in surrounding pixels so that shape of the target can be recovered for identification. Super resolution mapping is one such method for target identification and enhancement. Aim of this paper is to perform a comparative assessment of various existing super resolution mapping techniques and to present a super resolution mapping technique which can preferably work on non - random allocation of sub-pixels and non recursive optimization.