Advances in Hyperspectral Image Processing Techniques

12Fusion of Band Selection Methods for Hyperspectral Imaging

Yulei Wang¹, Lin Wang², and Chein-I Chang^1,3

¹Center for Hyperspectral Imaging in Remote Sensing (CHIRS), School of Information and Technology, Dalian Maritime University, Dalian, China

²School of Physics and Optoelectronic Engineering, Xidian University, Xi’an, China

³Remote Sensing Signal and Image Processing Laboratory, Department of Computer Science and Electrical Engineering, University of Maryland Baltimore County, Baltimore, MD, USA

12.1 Introduction

Hyperspectral imaging has emerged as a promising technique in remote sensing [1] due to its use of hundreds of contiguous spectral bands. However, it is also traded for an issue of how to effectively utilize such a wealth of spectral information. In various applications, such as classification, target detection, spectral unmixing, endmember finding/extraction, and so on, the spectral properties of material substances of interest may respond to different ranges of wavelengths. In this case, not all spectral bands are useful. Consequently, it is crucial to find appropriate wavelength ranges for particular applications of interest. This leads to a need of BS, which has become increasingly important in hyperspectral data exploitation.

Over the past years, many BS methods have been developed and reported in the literature [1–32]. In general, it can be categorized into two groups based on application‐independent and application‐dependent approaches. One is made up of ...