A background refinement method based on local density for hyperspectral anomaly detection

For anomaly detection, anomalies existing in the background will affect the detection performance. Accordingly, a background refinement method based on the local density is proposed to remove the anomalies from the background. In this work, the local density is measured by its spectral neighbors through a certain radius which is obtained by calculating the mean median of the distance matrix. Further, a two-step segmentation strategy is designed. The first segmentation step divides the original background into two subsets, a large subset composed by background pixels and a small subset containing both background pixels and anomalies. The second segmentation step employing Otsu method with an aim to obtain a discrimination threshold is conducted on the small subset. Then the pixels whose local densities are lower than the threshold are removed. Finally, to validate the effectiveness of the proposed method, it combines Reed-Xiaoli detector and collaborative-representation-based detector to detect anomalies. Experiments are conducted on two real hyperspectral datasets. Results show that the proposed method achieves better detection performance.摘要在高光谱异常检测的基于背景的估计的研究中, 背景中存在的异常像元会对背景估计的准确性产生不利影响, 通过获得一个不包含异常目标的纯净背景可以有助于提高异常目标检测算法的检测率。 因此, 本文提出一种基于局部密度的背景纯化方法以去除背景中的异常像元。经典的 RX 算法 (Reed-Xiaoli detector, RXD) 在背景统计信息估计时, 因初始背景中存在异常像元而产生失真现象。 这使 RX 算法的检测结果具有较高的虚警率。 在本文中, 将像元的局部密度作为该像元的异常度标签, 通过最大类间方差法去除初始背景中潜在的异常像元, 最后可以得到更纯净的背景集。 利用纯化背景得到更准确的背景统计信息, 可以有效降低 RX 算法的虚警率。首先, 通过度量一定半径内的光谱邻域, 获得像元的局部密度, 半径选取为距离矩阵的平均中位数; 然后, 基于局部密度设计了两步分割策略: 第一步, 将原始背景分割为一个包含大部分背景像元的集合及一个混合着背景像元和异常像元的小子集。第二步, 利用最大类间方差法处理小子集, 获得一个判别阈值, 去除局部密度小于该阈值的像元, 得到纯净的背景。 最后, 利用 RX 检测算法和基于协同表示的检测算法验证所提背景纯化算法的有效性。本文算法对于圣地亚哥海军基地数据和 SpecTIR 数据, AUC 值比 LRXD 分别提高了 0.0246 和 0.0068, 并且相比于其它算法有较高的检测率。 实验结果说明, 密度背景纯化方法有效抑制了异常点对背景数据的干扰, 使背景协方差和均值等参数更准确, 降低了虚警率。 在计算密度时, 需要人为设定半径值。 对于两幅高光谱数据, 在不同半径选择下 DBRAD 最大 AUC 差值为 0.0088 和 0.0012, 实验结果表明, 人为选择半径值可能会对最后的检测效果造成较大影响。 如何自适应选取半径将是下一步研究的重点。

[1]  Heesung Kwon,et al.  Kernel RX-algorithm: a nonlinear anomaly detector for hyperspectral imagery , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[2]  Lianru Gao,et al.  Probabilistic anomaly detector for remotely sensed hyperspectral data , 2014 .

[3]  Bo Du,et al.  Random-Selection-Based Anomaly Detector for Hyperspectral Imagery , 2011, IEEE Transactions on Geoscience and Remote Sensing.

[4]  Amit Banerjee,et al.  A support vector method for anomaly detection in hyperspectral imagery , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[5]  A KERNEL WEIGHTED RX ALGORITHM FOR ANOMALY DETECTION IN HYPERSPECTRAL IMAGERY: A KERNEL WEIGHTED RX ALGORITHM FOR ANOMALY DETECTION IN HYPERSPECTRAL IMAGERY , 2011 .

[6]  James E. Fowler,et al.  Nearest Regularized Subspace for Hyperspectral Classification , 2014, IEEE Transactions on Geoscience and Remote Sensing.

[7]  Paul E. Lewis,et al.  Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XV , 2006 .

[8]  Saeid Homayouni,et al.  Anomaly Detection in Hyperspectral Images Based on an Adaptive Support Vector Method , 2011, IEEE Geoscience and Remote Sensing Letters.

[9]  Qian Du,et al.  Collaborative Representation for Hyperspectral Anomaly Detection , 2015, IEEE Transactions on Geoscience and Remote Sensing.

[10]  Huijie Zhao,et al.  Local density-based anomaly detection in hyperspectral image , 2015 .

[11]  Peter Bajorski,et al.  Target Detection Under Misspecified Models in Hyperspectral Images , 2012, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[12]  A. Hadi,et al.  BACON: blocked adaptive computationally efficient outlier nominators , 2000 .

[13]  Xiaoli Yu,et al.  Adaptive multiple-band CFAR detection of an optical pattern with unknown spectral distribution , 1990, IEEE Trans. Acoust. Speech Signal Process..

[14]  Antonio J. Plaza,et al.  Analysis and Optimizations of Global and Local Versions of the RX Algorithm for Anomaly Detection in Hyperspectral Data , 2013, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[15]  Daniel R. Fuhrmann,et al.  A CFAR adaptive matched filter detector , 1992 .

[16]  Antonio J. Plaza,et al.  Dimensionality reduction and classification of hyperspectral image data using sequences of extended morphological transformations , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[17]  Inmaculada García,et al.  Anomaly detection based on a parallel kernel RX algorithm for multicore platforms , 2012 .

[18]  Dimitris G. Manolakis,et al.  Detection algorithms for hyperspectral imaging applications , 2002, IEEE Signal Process. Mag..

[19]  Andrew Chan,et al.  An SVDD-Based Algorithm for Target Detection in Hyperspectral Imagery , 2011, IEEE Geoscience and Remote Sensing Letters.

[20]  E. M. Winter,et al.  Anomaly detection from hyperspectral imagery , 2002, IEEE Signal Process. Mag..

[21]  Trac D. Tran,et al.  Simultaneous Joint Sparsity Model for Target Detection in Hyperspectral Imagery , 2011, IEEE Geoscience and Remote Sensing Letters.

[22]  Qian Du,et al.  Comparison between constrained energy minimization based approaches for hyperspectral imagery , 2003, IEEE Workshop on Advances in Techniques for Analysis of Remotely Sensed Data, 2003.

[23]  Qian Du,et al.  Unsupervised nearest regularized subspace for anomaly detection in hyperspectral imagery , 2013, 2013 IEEE International Geoscience and Remote Sensing Symposium - IGARSS.

[24]  Chein-I Chang,et al.  Anomaly detection and classification for hyperspectral imagery , 2002, IEEE Trans. Geosci. Remote. Sens..