多窗口融合判别子空间的高光谱图像异常检测

doi:10.3969/j.issn.0255-8297.2019.01.007

Abstract

Abstract:

Disturbed by heterogeneous background and noise, the direct application of traditional RX anomaly detection algorithm for hyperspectral image often results in high false alarms. In order to solve this problem, an improved RX algorithm based on discriminant subspace combined with multi-window fusion is proposed. Firstly, the discriminant features of dominant clustering samples are extracted. Secondly, the orthogonality subspace projection which is built by dominant feature vectors is used to obtain the maximum separation of the background and the target information, achieving the suppression of the background. Then multiple RX with different local window size are applied for the anomaly target enhanced data. Finally, the multi-window RX results are added together. The performance on testing methods is evaluated by AUC. The AUC statistical values of the NUANCE and HYDICE hyperspectral data anomaly detection experiments show that the multi-window fusion algorithm outperforms the classical global and local RX algorithms in detection performance, and it has a stronger inhibition on the background and noise, the detected abnormal target is more accurate, which proves the effectivity and feasibility of the proposed algorithm.

Key words: multi-window fusion, hyperspectral image, discriminant subspace, RX algorithm

CLC Number:

TP751

MA Chun-xiao, HUANG Yuan-cheng, HU Rong-ming, ZHANG Chun-sen. Discriminant Subspace and Multi-window Fusion RX Algorithm for Hyperspectral Image Anomaly Detection[J]. Journal of Applied Sciences, 2019, 37(1): 64-72.

References

[1] 李恪,姚崇斌,徐红新,谢宝蓉,尚吉扬. 一种改进的星上高光谱异常检测算法研究[J]. 上海航天,2017, 34(3):47-51. Li K, Yao C B, Xu H X, Xie B R, Shang J Y. Study on improved algorithm of hyperspectral anomaly detection on satellite[J]. Aerospace Shanghai, 2017, 34(3):47-51. (in Chinese)
[2] Zhao C H, Wang X P, Yao X F, Tian M H. A background refinement method based on local density for hyperspectral anomaly detection[J]. Journal of Central South University, 2018, 25(1):84-94.
[3] 赵春晖,王佳,王玉磊. 采用背景抑制和自适应阈值分割的高光谱异常目标检测[J]. 哈尔滨工程大学学报,2016, 37(2):278-283. Zhao C H, Wang J, Wang Y L. Hyperspectral anomaly detection based on background suppression and adaptive threshold segmentation[J]. Journal of Harbin Engineering University, 2016, 37(2):278-283. (in Chinese)
[4] 刘志刚,卢云龙,刘翔. 一种改进的高光谱RX异常检测方法[J]. 兵器装备工程学报,2014, 35(5):93-96. Liu Z G, Lu Y L, Liu X. A method for anomaly detection based on improved RX algorithm in hyperspectral imagery[J]. Journal of Sichuan Ordnance, 2014, 35(5):93-96. (in Chinese)
[5] 成宝芝,赵春晖,王玉磊. 结合光谱解混的高光谱图像异常目标检测SVDD算法[J]. 应用科学学报,2012, 30(1):82-88. Cheng B Z, Zhao C H, Wang Y L. SVDD algorithm with spectral unmixing for anomaly detection in hyperspectral images[J]. Journal of Applied Sciences, 2012, 30(1):82-88. (in Chinese)
[6] 赵春晖,姚淅峰. 基于局部核RX算法的高光谱实时检测[J]. 红外与毫米波学报,2016,35(6):708-714. Zhao C H, Yao X F. Local kernel RX algorithm-based hyperspectral real-time detection[J]. Journal of Infrared and Millimeter Waves, 2016, 35(6):708-714. (in Chinese)
[7] Li W, Du Q. Decision fusion for dual-window-based hyperspectral anomaly detector[J]. Journal of Applied Remote Sensing, 2015, 9(1):097297.
[8] 李晓慧. 聚类算法在高光谱图像异常目标检测中的应用研究[D]. 哈尔滨:哈尔滨工程大学,2012.
[9] 何高攀,杨桄,张筱晗,黄俊华,孟强强. 基于端元提取的高光谱异常目标检测[J]. 红外技术,2015, 37(10):836-841. He G P, Yang G, Zhang X H, Huang J H, Meng Q Q. Anomaly detection algorithm based on endmember extraction in hyperspectral imagery[J]. Infrared Technology, 2015, 37(10):836-841. (in Chinese)
[10] 李卫平,沈海斌. 基于接近函数的线性判别分析算法研究[J]. 电子技术,2017, 46(2):5-7. Li W P, Shen H B. Improved linear discriminant analysis algorithm based on close function[J]. Electronic Technology, 2017, 46(2):5-7. (in Chinese)
[11] 刘忠宝,王士同. 改进的线性判别分析算法[J]. 计算机应用,2011,31(1):250-253. Liu Z B, Wang S T. Improved linear discriminant analysis method[J]. Journal of Computer Applications, 2011, 31(1):250-253. (in Chinese)
[12] 赵春晖,李晓慧,王玉磊. 高光谱图像异常目标检测研究进展[J]. 电子测量与仪器学报, 2014, 28(8):803-811. Zhao C H, Li X H, Wang Y L. Research advance on anomaly detection for hyperspectral imagery[J]. Journal of Electronic Measurement and Instrument, 2014, 28(8):803-811. (in Chinese)
[13] 梅锋,赵春晖. 基于空域滤波的核RX高光谱图像异常检测算法[J]. 哈尔滨工程大学学报,2009, 30(6):697-702. Mei F, Zhao C H. Spatial filter based anomaly detection algorithm for hyperspectral imagery kernel RX detectors[J]. Journal of Harbin Engineering University, 2009, 30(6):697-702. (in Chinese)
[14] 黄远程,钟燕飞,赵野鹤,朱卫恒. 联合盲分解与稀疏表达的高光谱图像异常目标检测[J]. 武汉大学学报(信息科学版),2015, 40(9):1144-1150. Huang Y C, Zhong Y F, Zhao Y H, Zhu W H. Joint blind unmixing and sparse representation for anomaly detection in hyperspectral image[J]. Geomatics and Information Science of Wuhan University (Information Science Edition), 2015, 40(9):1144-1150. (in Chinese)
[15] 程凯. 无先验信息的高光谱图像小目标检测算法研究[D]. 苏州:苏州大学,2015.
[16] 蒋琳,谭晓阳,刘俊. 稀疏低秩双线性判别模型及其应用[J]. 计算机科学与探索,2010, 4(7):654-661. Jiang L, Tan X Y, Liu J. Sparse low rank bilinear discriminative model and its application[J]. Journal of Frontiers of Computer Science & Technology, 2010, 4(7):654-661. (in Chinese)
[17] Li J Y, Zhang H Y, Zhang L P, Ma L. Hyperspectral anomaly detection by the use of background joint sparse representation[J]. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2015, 8(6):2523-2533.

Discriminant Subspace and Multi-window Fusion RX Algorithm for Hyperspectral Image Anomaly Detection

PDF

Knowledge

Abstract

Cite this article

share this article

References

Related Articles 3

Recommended Articles

Metrics

Comments

[1]	LIU Yang, XU Hong-ping, YI Hang, SHI Qing-ping, XIA Wei-qiang, KANG Jian. Hyperspectral and Multi-spectral Data Fusion Based on Constraint CNMF [J]. Journal of Applied Sciences, 2016, 34(6): 651-660.
[2]	JIA Ying-biao, FENG Yan, YUAN Xiao-ling, WEI Jiang. Block Compressed Sensing Sampling and Reconstruction Using Spectral Prediction for Hyperspectral Images [J]. Journal of Applied Sciences, 2014, 32(3): 281-286.
[3]	WANG Li-guo, LIU Dan-feng, ZHAO Liang. Distance-Preserving Color Visualization Model for Hyperspectral Imagery [J]. Journal of Applied Sciences, 2013, 31(1): 72-78.