收稿：2017-03-29，
录用：2017-9-25，
纸质出版：2019-05
移动端阅览
高光谱图像异常点检测中，传统RX异常点检测算法忽略了空间相关性，背景估计不准确。本文提出了一种基于图像局部邻域光谱显著性分析的加权RX算法。该算法通过引入图像显著性分析，对基于概率密度为权重的图像背景建模进行改进，建立光谱显著性权重图，重新定义RX算法中的均值向量和协方差矩阵，并给不同的目标赋予不同的权值，达到优化背景估计的目的。利用合成高光谱数据和真实高光谱数据进行异常点检测实验，结果表明，对于同一组数据，本文算法检测到的异常点数比传统算法多，虚警率较低，有效地提高了检测率。
With the development of spectral imaging technique and its data processing technology
anomaly detection using hyperspectral data has become a popular topic. Anomaly detection refers to the search for sparse pixels of unknown spectral signals in hyperspectral imagery. Given that the anomaly detection is unsupervised
providing a priori information is necessary. Thus
anomaly detection has a strong practicality. Considering the lack of spatial correlation and low normal distribution adaptation
the traditional RX algorithm has an inaccurate background estimation. Thus
this algorithm is unsuitable for detecting hyperspectral data. In this study
a saliency weighted RX algorithm is proposed on the basis of the local neighborhood spectra of an image. When the human eye observes an image
the first object that is viewed is frequently the most significant. The significance of the saliency detection algorithm is to identify this goal. The saliency map is a 2D image of the same size as the original image to represent the significance of the corresponding pixel in the original image. In this algorithm
the image background modeling based on probability density is improved by introducing a saliency analysis method. Afterward
the spectral saliency map is established
and the mean vector and covariance matrix of the RX algorithm are redefined. Saliency weighted RX algorithm provides different weights to optimize the background estimation. Anomaly detection experiments are conducted using synthetic and real hyperspectral data. Synthetic data experimental results show that
for each target
the number of anomalies detected using the saliency weighted RX algorithm is more than that of the traditional algorithms
and the saliency weighted RX algorithm can detect anomalies with abundance below 0.1. By contrast
traditional algorithms cannot detect these anomalies. Moreover
the false alarm pixels of the traditional algorithms are distributed in various positions
whereas the saliency weighted RX algorithm concentrates on an area called a false alarm area. This area can be removed effectively by morphological filtering. Real data experimental results show that the saliency weighted RX algorithm corresponds to the largest AUC value and has the optimal detection results. The traditional RX algorithm assumes that the background model follows a multivariate Gaussian distribution and does not perform well in hyperspectral imagery. The method of saliency analysis in the field of computer vision can be effectively analyzed in the spatial domain. This phenomenon compensates for the shortcomings of the RX algorithm to ignore spatial correlation
thus detecting the anomalies synchronized in the spatial and spectral domains. The saliency weighted RX algorithm uses a saliency analysis method to provide the background and anomaly pixels with a different weight
thereby improving the adaptability of the background model. Through the experiment of synthetic and real data
the saliency weighted algorithm can improve the detection probability while reducing the false alarm rate in comparison with the traditional RX algorithm and has a certain anti-noise ability.
Ashton E A and Schaum A. 1998. Algorithms for the detection of sub-pixel targets in multispectral imagery. Photogrammetric Engineering and Remote Sensing, 64(7): 723–731
Chang C I and Chiang S S. 2002. Anomaly detection and classification for hyperspectral imagery. IEEE Transactions on Geoscience and Remote Sensing, 40(6): 1314–1325
Chang C I, Li Y, Hobbs M C, Schultz R C and Liu W M. 2015. Progressive band processing of anomaly detection in hyperspectral imagery. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 8(7): 3558–3571
高昆, 刘莹, 王丽静, 朱振宇, 程灏波. 2015. 基于高斯马尔科夫模型的高光谱异常目标检测算法研究. 光谱学与光谱分析, 35(10): 2846–2850
Gao K, Liu Y, Wang L J, Zhu Z Y and Cheng H B. 2015. A hyperspectral imagery anomaly detection algorithm based on gauss-Markov model. Spectroscopy and Spectral Analysis, 35(10): 2846–2850 (
Goferman S, Zelnik-Manor L, Tal A. 2012. Context-aware saliency detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 34(10): 1915–1926
Guo Q D, Zhang B, Qiong R, Gao L R, Li J and Plaza A. 2014. Weighted-RXD and linear filter-based RXD: improving background statistics estimation for anomaly detection in hyperspectral imagery. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(6): 2351–2366
Itti L, Koch C, Niebur E. 1998. A model of saliency-based visual attention for rapid scene analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 20(11): 1254–1259
Khazai S, Homayouni S, Safari A and Mojaradi B. 2011. Anomaly detection in hyperspectral images based on an adaptive support vector method. IEEE Geoscience and Remote Sensing Letters, 8(4): 646–650
李君浩, 刘志. 2015. 基于视觉显著性图与似物性的对象检测. 计算机应用, 35(12): 3560–3564
Li J H and Liu Z. 2015. Object detection based on visual saliency map and objectness. Journal of Computer Applications, 35(12): 3560–3564 (
Liu W M and Chang C I. 2013. Multiple-window anomaly detection for hyperspectral imagery. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 6(2): 644–658
Molero J M, Garzón E M, García I and Plaza A. 2013. Analysis and optimizations of global and local versions of the RX algorithm for anomaly detection in hyperspectral data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 6(2): 801–814
Reed I S and Yu X. 1990. Adaptive multiple-band CFAR detection of an optical pattern with unknown spectral distribution. IEEE Transactions on Acoustics, Speech, and Signal Processing, 38(10): 1760–1770
Stellman C M, Hazel G G, Bucholtz F, Michalowicz J V, Stocker A D and Schaaf W. 2000. Real-time hyperspectral detection and cuing. Optical Engineering, 39(7): 1928–1935
童庆禧, 张兵, 郑兰芬. 2006. 高光谱遥感: 原理、技术与应用. 北京: 高等教育出版社: 40–47
Tong Q X, Zhang B and Zheng L F. 2006. Hyperspectral Remote Sensing: Principle, Technology and Application. Beijing: Higher Education Press: 40–47
童庆禧, 张兵, 张立福. 2016. 中国高光谱遥感的前沿进展. 遥感学报, 20(5): 689–707
Tong Q X, Zhang B and Zhang L F. 2016. Current progress of hyperspectral remote sensing in China. Journal of Remote Sensing, 20(5): 689–707 (
Xu Y, Wu Z B, Li J, Plaza A and Wei Z H. 2016. Anomaly detection in hyperspectral images based on low-rank and sparse representation. IEEE Transactions on Geoscience and Remote Sensing, 54(4): 1990–2000
Yu X, Reed I S and Stocker A D. 1993. Comparative performance analysis of adaptive multispectral detectors. IEEE Transactions on Signal Processing, 41(8): 2639–2656
Yuan Y, Wang Q and Zhu G K. 2015. Fast hyperspectral anomaly detection via high-order 2-D crossing filter. IEEE Transactions on Geoscience and Remote Sensing, 53(2): 620–630
张兵. 2016. 高光谱图像处理与信息提取前沿. 遥感学报, 20(5): 1062–1090
Zhang B. 2016. Advancement of hyperspectral image processing and information extraction. Journal of Remote Sensing, 20(5): 1062–1090 (
赵春晖, 王立国, 齐滨. 2016. 高光谱遥感图像处理方法及应用. 北京: 电子工业出版社: 261–271
Zhao C H, Wang L G and Qi B. 2016. Hyperspectral Remote Sensing Images Processing Methods and Applications. Beijing: Publishing House of Electronics Industry: 261–271
赵春晖, 姚淅峰. 2016. 基于局部核RX算法的高光谱实时检测. 红外与毫米波学报, 35(6): 708–714
Zhao C H, Yao X F. 2016. Local kernel RX algorithm-based hyperspectral real-time detection. Journal of Infrared and Millimeter Waves, 35(6): 708–714 (
Zhao R, Du B and Zhang L P. 2014. A robust nonlinear hyperspectral anomaly detection approach. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(4): 1227–1234
相关文章
相关作者
相关机构
京公网安备11010802024621
