Hyperspectral image classification in compressively sensed band domain

Despite that hyperspectral technology has continued to improve over the years, it still has practical constraints on size, weight, and power (SWaP). One major issue is its use of a large number of very fine spectral bands which creates challenges in both data archival and processing. Compressive sensing (CS) becomes an enabling technology that can be used to reduce the overall data processing and SWaP requirements. This chapter explores the viability of performing classification for hyperspectral data on a compressively sensed band domain (CSBD) via CS instead of operating the original data space (ODS). In particular, the well-known restricted isometry property (RIP) and a random spectral sampling strategy are investigated for hyperspectral image classification (HSIC) in CSBD. A mathematical analysis is also presented to show that the classification error can be expressed in terms of the restricted isometry constant (RIC) so that the HSIC in ODS can be achieved in CSBD provided that sufficient band sensing conditions are met. To validate the proposed CS-HSIC, a set of hyperspectral real experiments are performed where a commonly used spectral-spatial classification algorithm based on support vector machine (SVM) and edge-preserving filters (EPFs) is implemented for a comparative study and analysis. The results clearly demonstrate the potential of CS-HSIC in future research directions.