Hyperspectral Image Compression and Reconstruction Based on Block-Sparse Dictionary Learning
- PDF / 2,659,491 Bytes
- 16 Pages / 595.276 x 790.866 pts Page_size
- 96 Downloads / 237 Views
RESEARCH ARTICLE
Hyperspectral Image Compression and Reconstruction Based on BlockSparse Dictionary Learning Yanwen Chong1
•
Weiling Zheng1 • Haonan Li1 • Zhixi Qiao2 • Shaoming Pan1
Received: 7 January 2017 / Accepted: 3 June 2018 Ó Indian Society of Remote Sensing 2018
Abstract A large amount of hyperspectral image (HSI) data poses a significant challenge for transmission and storage. A new signal processing mechanism—compressed sensing (CS)—is appropriate for processing signals with a massive amount of data and can achieve high reconstruction accuracy. According to the structural properties of HSI, the same ground features show the same spectral properties. In this paper, an approach is proposed to compress and reconstruct HSI based on CS and block-sparse dictionary learning. Primarily, a dictionary of a given set of signal is trained and prior knowledge is not required on the association of the training dataset into groups. Then, a measurement matrix is used to compress an HSI cube to reduce the data volume of the signal. Finally, we use the trained block-sparse dictionary to reconstruct the image, along with the HSI feature classification information. Our experimental results showed that, for block-sparse HSI data, the proposed approach significantly improved the performance compared with other related state of the art methods. Keywords Hyperspectral image (HSI) compression and reconstruction Measurement matrix Block-sparse dictionary
Introduction A hyperspectral image (HSI) is a three-dimensional cube signal, including two-dimensional spatial information and one-dimensional spectral information. Each band of HSI corresponds to a two-dimensional image. The pixels at the same position of different spectral bands constitute a spectral curve. Because different ground features have & Shaoming Pan [email protected] Yanwen Chong [email protected] Weiling Zheng [email protected] Haonan Li [email protected] Zhixi Qiao [email protected] 1
State Key Laboratory for Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, 129 Luoyu Road, Wuhan 430079, China
2
College of Remote Sensing and Information Engineering, Wuhan University, 129 Luoyu Road, Wuhan 430079, China
different spectral properties, the spectral curve of each band is not the same. Therefore, HSI is widely applied in environmental and military detection, agricultural planning, mineral exploration, and other fields (Smith et al. 2001; Chan, et al. 2011; Toivanen et al. 2005a, b). To accurately identify the different ground features, the higher spectral and spatial resolution of HSI is required (Shaw and Burke 2003). The spectral resolution of HSI reflects the number of bands, and the spatial resolution indicates the size of the smallest object that can be identified. In recent years, with the advancement of remote sensing imaging technology, the spectral and spatial resolution of remote sensing images has significantly increased, and the HSI data volume has thus dramatically increased. The HSI data contains a larg
Data Loading...
