Advances in Hyperspectral Image Processing Techniques

15Swarm Intelligence Optimization‐Based Spectral Unmixing

Lianru Gao¹, Xu Sun¹, Zhu Han^1,2, Lina Zhuang³, Wenfei Luo⁴, and Bing Zhang^1,2

¹Key Laboratory of Digital Earth Science, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, China

²College of Resources and Environment, University of Chinese Academy of Sciences, Beijing, China

³Department of Mathematics, Hong Kong Baptist University, Hong Kong, China

⁴School of Geographical Science, South China Normal University, Guangzhou, China

15.1 Introduction

Mixed pixels are common in hyperspectral image due to relatively low spatial resolution. The main purpose of spectral unmixing is to analyze the materials (called endmembers) contained in the mixed pixel and their proportions (called abundances). The basic models normally adopted in spectral unmixing are LMM, NLMM, and NCM. Among them, LMM is widely used in mixed pixel problems, because it can satisfy the physical principles of the spectral mixing process under certain conditions, and the form of LMM is simple. The endmember extraction algorithm based on LMM can be further divided into two main categories: geometrical and statistical. Of these approaches, geometrical methods have the longest history of study, and these algorithms are the most abundant. Geometrical methods are based on convex geometry concepts. According to different assumptions, the geometrical methods can be classified in two types: pure pixel‐based methods and minimum volume‐based ...