Advances in Hyperspectral Image Processing Techniques

4Constrained Band Selection for Hyperspectral Imaging

Chein‐I Chang

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

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

4.1 Introduction

With significantly refined spectral resolution, a hyperspectral image (HSI) generally contains abundant spectral information that can be used to uncover many subtle material substances present in a single pixel at subscale. As a consequence, HSI has emerged as an advanced remote sensing technique that can resolve many issues which cannot be solved by multispectral imaging. In many practical applications, HSI has found its great potential in data exploitation, such as missile/vehicle detection in combat, rare materials in geology, pesticide residual detection in food safety and inspection, soil contamination in agriculture, agriculture, drug traffic in law enforcement, rescue and search in disaster area, and water/toxic pollution for environmental monitoring. While HSI has proved to be a very valuable remote sensing technology, it also suffers from so‐called curse of dimensionality. In order to mitigate this dilemma, data dimensionality reduction (DR) is generally implemented to reduce data volume prior to data processing [1]. In general, two approaches to DR are of particular interest. ...