Hyperspectral Data Processing: Algorithm Design and Analysis

21.6 Hyperspectral Compresssion by PBDP

In analogy with PSDP used to design progressive spectral dimensionality reduction (PSDR) via DP and progressive spectral dimensionality expansion (PSDE) via DP in Chapter 20 PBDP is also a backbone of two major dual processes, progressive band dimensionality reduction (PBDR) via BP, and progressive band dimensionality expansion (PBDE) via BP. It has been shown in Chang et al. (2010) that the number of target signature substances of interest in hyperspectral imagery was generally estimated between n_VD and 2n_VD. So, if we use one spectral band to accommodate a specific material substance, then the number of bands required to be selected must be equal to or greater than the number of signature substances, which is determined by VD. So, these two numbers, n_VD and 2n_VD can be used to provide a reasonable lower bound and an upper bound on the value of for PBDE and PBDR, respectively.

21.6.1 Progressive Band Dimensionality Reduction Via BP

PBDR is a process that allows users to reduce a large number of spectral bands by removing certain spectral bands with low priorities. It starts with a maximal number of spectral bands and begins to eliminate a number of spectral bands with lowest priorities from the currently processed band set until the performance of data processing is not satisfied or it reaches the minimal number of spectral bands, which can ...