Limits...
A Weighted Spatial-Spectral Kernel RX Algorithm and Efficient Implementation on GPUs

View Article: PubMed Central - PubMed

ABSTRACT

The kernel RX (KRX) detector proposed by Kwon and Nasrabadi exploits a kernel function to obtain a better detection performance. However, it still has two limits that can be improved. On the one hand, reasonable integration of spatial-spectral information can be used to further improve its detection accuracy. On the other hand, parallel computing can be used to reduce the processing time in available KRX detectors. Accordingly, this paper presents a novel weighted spatial-spectral kernel RX (WSSKRX) detector and its parallel implementation on graphics processing units (GPUs). The WSSKRX utilizes the spatial neighborhood resources to reconstruct the testing pixels by introducing a spectral factor and a spatial window, thereby effectively reducing the interference of background noise. Then, the kernel function is redesigned as a mapping trick in a KRX detector to implement the anomaly detection. In addition, a powerful architecture based on the GPU technique is designed to accelerate WSSKRX. To substantiate the performance of the proposed algorithm, both synthetic and real data are conducted for experiments.

No MeSH data available.


The AUC of WSSKRX with the changing window size . (a) Synthetic dataset; (b) San Diego Airport dataset; (c) SpecTIR dataset.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5375727&req=5

sensors-17-00441-f012: The AUC of WSSKRX with the changing window size . (a) Synthetic dataset; (b) San Diego Airport dataset; (c) SpecTIR dataset.

Mentions: Figure 12 gives the AUC curves of WSSKRX using different size of dual windows on three datasets, respectively. When the size of windows on synthetic dataset, San Diego Airport dataset and SpecTIR dataset are (3, 11), (5, 11) and (3, 11) respectively, the detection performance of WSSKRX is optimal. The optimal inner window size win used to protect anomaly information is also larger, as the anomaly target of the San Diego Airport dataset is larger than other two datasets. The window sizes in general depend on the size of target pixel in the image and local-double-window model is more suitable for small target data processing. Generally, the analysis of window sizes were considered in the range from (3, 11) to (7, 15) in the field of anomaly detection. For the RBF kernel parameter and the weighting factor, we can determine a proper parameter size by cross-validation.


A Weighted Spatial-Spectral Kernel RX Algorithm and Efficient Implementation on GPUs
The AUC of WSSKRX with the changing window size . (a) Synthetic dataset; (b) San Diego Airport dataset; (c) SpecTIR dataset.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5375727&req=5

sensors-17-00441-f012: The AUC of WSSKRX with the changing window size . (a) Synthetic dataset; (b) San Diego Airport dataset; (c) SpecTIR dataset.
Mentions: Figure 12 gives the AUC curves of WSSKRX using different size of dual windows on three datasets, respectively. When the size of windows on synthetic dataset, San Diego Airport dataset and SpecTIR dataset are (3, 11), (5, 11) and (3, 11) respectively, the detection performance of WSSKRX is optimal. The optimal inner window size win used to protect anomaly information is also larger, as the anomaly target of the San Diego Airport dataset is larger than other two datasets. The window sizes in general depend on the size of target pixel in the image and local-double-window model is more suitable for small target data processing. Generally, the analysis of window sizes were considered in the range from (3, 11) to (7, 15) in the field of anomaly detection. For the RBF kernel parameter and the weighting factor, we can determine a proper parameter size by cross-validation.

View Article: PubMed Central - PubMed

ABSTRACT

The kernel RX (KRX) detector proposed by Kwon and Nasrabadi exploits a kernel function to obtain a better detection performance. However, it still has two limits that can be improved. On the one hand, reasonable integration of spatial-spectral information can be used to further improve its detection accuracy. On the other hand, parallel computing can be used to reduce the processing time in available KRX detectors. Accordingly, this paper presents a novel weighted spatial-spectral kernel RX (WSSKRX) detector and its parallel implementation on graphics processing units (GPUs). The WSSKRX utilizes the spatial neighborhood resources to reconstruct the testing pixels by introducing a spectral factor and a spatial window, thereby effectively reducing the interference of background noise. Then, the kernel function is redesigned as a mapping trick in a KRX detector to implement the anomaly detection. In addition, a powerful architecture based on the GPU technique is designed to accelerate WSSKRX. To substantiate the performance of the proposed algorithm, both synthetic and real data are conducted for experiments.

No MeSH data available.