Modified Kernel RX Algorithm Based on Background Purification and Inverse-of-Matrix-Free Calculation

The kernel RX detector (KRXD) has better performance than the RX algorithm in anomaly detection (AD). However, it generally suffers from two challenges: 1) it is more prone to background contamination by anomalous pixels and noise in local statistics since the local AD is normally implemented for KRXD to relieve high computational complexity in global AD and 2) the inverse of the kernelized background covariance matrix is usually rank deficient. Accordingly, this letter proposes a Gaussian background purification approach according to background data samples probability distribution and an inverse-of-matrix-free method based on kernel PCA to address the above problems, respectively. The experimental results indicate that the improved KRXD overcomes both the difficulties and procures preferable effects.

[1]  Antonio J. Plaza,et al.  Weighted-RXD and Linear Filter-Based RXD: Improving Background Statistics Estimation for Anomaly Detection in Hyperspectral Imagery , 2014, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[2]  Stanley R. Rotman,et al.  Anomaly detection in non-stationary backgrounds , 2010, 2010 2nd Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing.

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

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

[5]  S Matteoli,et al.  A tutorial overview of anomaly detection in hyperspectral images , 2010, IEEE Aerospace and Electronic Systems Magazine.

[6]  Weiyue Li,et al.  Low-rank and sparse matrix decomposition-based anomaly detection for hyperspectral imagery , 2014 .

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

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

[9]  Heesung Kwon,et al.  Adaptive anomaly detection using subspace separation for hyperspectral imagery , 2003 .

[10]  Alexander J. Smola,et al.  Learning with kernels , 1998 .

[11]  Bo Du,et al.  A Low-Rank and Sparse Matrix Decomposition-Based Mahalanobis Distance Method for Hyperspectral Anomaly Detection , 2016, IEEE Transactions on Geoscience and Remote Sensing.

[12]  M. Eismann.  Hyperspectral Remote Sensing , 2012 .

[13]  S.R. Rotman,et al.  Improved covariance matrices for point target detection in hyperspectral data , 2008, 2009 IEEE International Conference on Microwaves, Communications, Antennas and Electronics Systems.

[14]  James Theiler,et al.  Problematic Projection to the In-Sample Subspace for a Kernelized Anomaly Detector , 2016, IEEE Geoscience and Remote Sensing Letters.

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