TY - JOUR
T1 - Hyperspectral Mixed Noise Removal by ℓ1-Norm-Based Subspace Representation
AU - Zhuang, Lina
AU - Ng, Michael K.
N1 - Funding Information:
Manuscript received December 4, 2019; revised February 6, 2020; accepted March 1, 2020. Date of publication March 18, 2020. This work was supported in part by the Hong Kong Baptist University Start-up under Grant 21.4551.162562. The work of Ng was supported in part by the Hong Kong Research Grants Council General Research Fund under Grants 12306616, 12200317, 12300218, and 12300519. (Corresponding author: Lina Zhuang.) Lina Zhuang is with the Department of Mathematics, Hong Kong Baptist University (HKBU), Hong Kong (e-mail: [email protected]).
PY - 2020/3/18
Y1 - 2020/3/18
N2 - This article introduces a new hyperspectral image (HSI) denoising method that is able to cope with additive mixed noise, i.e., mixture of Gaussian noise, impulse noise, and stripes, which usually corrupt hyperspectral images in the acquisition process. The proposed method fully exploits a compact and sparse HSI representation based on its low-rank and self-similarity characteristics. In order to deal with mixed noise having a complex statistical distribution, we propose to use the robust ell _1 data fidelity instead of using the ell _2 data fidelity, which is commonly employed for Gaussian noise removal. In a series of experiments with simulated and real datasets, the proposed method competes with state-of-the-art methods, yielding better results for mixed noise removal.
AB - This article introduces a new hyperspectral image (HSI) denoising method that is able to cope with additive mixed noise, i.e., mixture of Gaussian noise, impulse noise, and stripes, which usually corrupt hyperspectral images in the acquisition process. The proposed method fully exploits a compact and sparse HSI representation based on its low-rank and self-similarity characteristics. In order to deal with mixed noise having a complex statistical distribution, we propose to use the robust ell _1 data fidelity instead of using the ell _2 data fidelity, which is commonly employed for Gaussian noise removal. In a series of experiments with simulated and real datasets, the proposed method competes with state-of-the-art methods, yielding better results for mixed noise removal.
KW - High-dimensional data
KW - hyperspectral destriping
KW - hyperspectral restoration
KW - low-rank representation
KW - nonlocal patch
KW - self-similarity
UR - http://www.scopus.com/inward/record.url?scp=85083452049&partnerID=8YFLogxK
U2 - 10.1109/JSTARS.2020.2979801
DO - 10.1109/JSTARS.2020.2979801
M3 - Journal article
AN - SCOPUS:85083452049
SN - 1939-1404
VL - 13
SP - 1143
EP - 1157
JO - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
JF - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
M1 - 9040508
ER -