Hyperspectral Data Processing: Algorithm Design and Analysis

26.6 Conclusions

This chapter introduces a new concept of PSSC for hyperspectral signature characterization. It is derived from a technique called MPCM that was previously developed for progressive image reconstruction and edge detection. Unlike the commonly used SSC that performs coding with hard decision, PSSC characterizes a hyperspectral signature in a sequence of soft decisions in multiple stages to produce a spectral profile of progressive changes in spectral variation of a spectral signature vector. The idea of MPCM-based PSSC (MPCM-PSSC) is to use a sequence of soft decision-based quantizers to generate priority codes for a hyperspectral signature vector that can be used to prioritize signature values across its spectral range whose priorities are specified by stage levels implemented in various stages. Such priority codes allow users to decompose and reconstruct a hyperspectral signature vector progressively in accordance with the priorities assigned to spectral signature values specified by various wavelengths. As a result, a spectral profile of progressive changes in spectral variation can be generated for a hyperspectral signature vector and can be further used to dictate subtle differences in spectral characterization. To substantiate the utility of MPCM-PSSC in applications spectral discrimination and identification are used for illustration. Experiments are also conducted to demonstrate unique features of MPCM-PSSC in hyperspectral signature characterization such ...