Recursive Hyperspectral Band Processing for Active Target Detection: Constrained Energy Minimization

Abstract

Chapter 5 extends constrained energy minimization (CEM) to a real-time processing version of CEM, called real-time CEM (RT CEM) that allows CEM to process data according to a band-interleaved-by-pixel/sample (BIP/BIS) data acquisition format sample by sample recursively in real time. This chapter extends CEM to another type of real-time implementation of CEM from a band-sequential (BSQ) format perspective to recursive hyperspectral band processing (RHBP) of CEM (RHBP-CEM) that can perform CEM according to BSQ band by band recursively in real time. In doing so we introduce a new concept, called causal band correlation matrix (CBCRM), which is a correlation matrix formed by only those bands that were already visited up to the band currently being processed but not bands yet to be visited in the future, to replace the global sample correlation matrix R so that CBCRM can be updated band by band in real time. The RHBP-CEM presented in this chapter allows CEM to perform target detection progressively and recursively whenever bands are available without waiting for the completion of band collection. With such an advantage RHBP-CEM has potential in data transmission and communication, specifically in satellite data processing.