Hyperspectral Data Processing: Algorithm Design and Analysis

24.5 Conclusions

Binary coding is a simple and effective technique to be used for spectral characterization. This chapter explores various ways to extend one of most widely used binary coding technique, SPAM, developed by Mazer et al. (1988) as well as a spectral feature-based coding method developed by Qian et al. (1996). Three new binary coding techniques, median partition, halfway partition, and equal probability partition are developed and can be further implemented in conjunction with SPAM and Qian et al.'s method to derive new binary coding methods. In order to effectively conduct performance analysis and evaluation between two coding methods, a new criterion, APDP, is also introduced for performance measure. With the help of the proposed APDP, a set of selected eight binary coding methods is evaluated for comparative study via computer simulations and real hyperspectral image experiments in spectral discrimination and identification. Experimental results demonstrate that each method has its own merit and can be used in different applications.