Advances in Hyperspectral Image Processing Techniques

10Hyperspectral and LiDAR Data Fusion

Qian Du¹, Wei Li², and Chiru Ge³

¹Department of Electrical and Computer Engineering, Mississippi State University, Starkville, MS, USA

²College of Information and Electronics, Beijing Institute of Technology, Beijing, China

³School of Information Science and Engineering, Shandong Normal University, Jinan, China

10.1 Introduction

Hyperspectral imaging (HSI) sensors acquire hundreds of narrow spectral bands with high spectral resolution, while laser scanning can measure surface topography and 3‐D characterization of targets. Extending laser‐based intensity into broadband spectral range using an active hyperspectral light detection and ranging (LiDAR) generates additional spectral features for target classification [1–5]. A more flexible strategy is to fuse data from a passive hyperspectral sensor and an active laser scanner [6]. The combined use of HSI and LiDAR results in higher classification accuracies than using each source separately. HSI and LiDAR have been used in combination in many applications successfully, such as biomass estimation [7, 8], vegetation monitoring [9, 10], tree species classification [11, 12], and improved bathymetry estimation [13].

Figure 10.1(a) shows an HSI, whose poor spatial contrast between objects limits its performance in spatial feature presentation. The LiDAR‐based digital surface model (DSM) shown in Figure 10.1(b) provides more accurate elevation information and useful spatial contrast. Thus, fusing ...