Hyperspectral Data Processing: Algorithm Design and Analysis

25

Vector Coding for Hyperspectral Signatures

Spectral signature coding is generally performed by encoding a spectral signature vector across its spectral coverage and using the Hamming distance to measure signature similarity as discussed in Chapter 24. The effectiveness of such a spectral signature coding largely relies on how well the Hamming distance can capture spectral variations that characterize a signature vector. Unfortunately, in most cases, such binary coding does not provide sufficient information for signature analysis due to inability of the Hamming distance in capturing the band-to-band variation, in which case the Hamming distance can be considered a memoryless distance. Therefore, one approach is to extend the Hamming distance to a distance with memory. This chapter introduces two new concepts, referred to as spectral derivative feature coding (SDFC) and spectral feature probabilistic coding (SFPC), for signature coding. SDFC is derived from texture features used in texture classification to dictate gradient changes among adjacent bands in characterizing spectral variations so as to improve spectral discrimination and classification. SFPC implements the well-known arithmetic coding (AC) in two different ways to encode a signature vector in a probabilistic manner, called circular-SFPC (C-SFPC) and split-SFPC (S-SFPC). The values resulting from AC are then used to measure the distance between two signature vectors. The experimental results show that these two signature ...