Hyperspectral Anomaly Detection with Harmonic Analysis and Low-Rank Decomposition

Xiang, Pei; Song, Jiangluqi; Li, Huan; Gu, Lin; Zhou, Huixin

doi:10.3390/rs11243028

Cited by 9 publications

(6 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To compare the detection performance of the model proposed and other compared algorithms, the receiver operating characteristic (ROC) and the area under the curve (AUC) value are adopted as the evaluation criteria [44,45]. In the HAD evaluation area, ROC is the most popular [46]. It is like a function where the true positive rate (TPR) is uniquely determined when the false positive rate (FPR) is fixed.…”

Section: Evaluation Criteriamentioning

confidence: 99%

Unsupervised Generative Adversarial Network with Background Enhancement and Irredundant Pooling for Hyperspectral Anomaly Detection

Shu

Wang

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

Lately, generative adversarial networks (GAN)-based methods have drawn extensive attention and achieved a promising performance in the field of hyperspectral anomaly detection (HAD) owing to GAN’s powerful data generation capability. However, without considering the background spatial features, most of these methods can not obtain a GAN with a strong background generation ability. Besides, they fail to address the hyperspectral image (HSI) redundant information disturbance problem in the anomaly detection part. To solve these issues, the unsupervised generative adversarial network with background spatial feature enhancement and irredundant pooling (BEGAIP) is proposed for HAD. To make better use of features, spatial and spectral features union extraction idea is also applied to the proposed model. To be specific, in spatial branch, a new background spatial feature enhancement way is proposed to get a data set containing relatively pure background information to train GAN and reconstruct a more vivid background image. In a spectral branch, irredundant pooling (IP) is invented to remove redundant information, which can also enhance the background spectral feature. Finally, the features obtained from the spectral and spatial branch are combined for HAD. The experimental results conducted on several HSI data sets display that the model proposed acquire a better performance than other relevant algorithms.

show abstract

Section: Evaluation Criteriamentioning

confidence: 99%

Unsupervised Generative Adversarial Network with Background Enhancement and Irredundant Pooling for Hyperspectral Anomaly Detection

Shu

Wang

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Two well-known methods, receiver operating characteristic (ROC) and area under curve value (AUC), are used to evaluate the performance of the anomaly detection algorithms. ROC [58] can reflect the relationship between detection rate (true positive rate, TPR) and false-positive rate (FPR), thus can comprehensively evaluate the effectiveness of the algorithm. The higher the curve, the better the algorithm performance.…”

Section: B Evaluation Metricsmentioning

confidence: 99%

Subfeature Ensemble-Based Hyperspectral Anomaly Detection Algorithm

Wang

Feng

Quan

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Hyperspectral images (HSIs) have always played an important role in remote sensing applications. Anomaly detection has become a hot spot in HSI processing in recent years. The popular detecting method is to accurately segment anomalies from the background. Informative bands are very important for the accuracy improvement of the detection technology. However, most of the abnormal targets segmentation methods focus on the usage of all the spectral features, thus are easily affected by redundant bands or feature noise. A hyperspectral anomaly detection algorithm based on sub-feature ensemble is proposed in this letter. The proposed method consists of the following steps. Firstly, the bands of the original HSI are normalized and randomly divided into several sub-feature sets according to different proportions. Secondly, six methods including the prior-based tensor approximation algorithm (PTA), Reed-Xiaoli method (RX), a low-rank and sparse representation method (LRASR), a low-rank and sparse matrix decomposition-based Mahalanobis distance method (LSMAD), the graph and total variation regularized low-rank representation-based method (GTVLRR), and a method based on tensor principal component analysis (TPCA) are applied to detect anomalies on the original HSI, and the method with the best performance is used to obtain an enhanced feature set. Then the enhanced features and the sub-features are ensembled iteratively to construct a new dataset. Finally, the PTA method is operated on the data set with ensemble features to get the final abnormal target results. Six hyperspectral data sets are used in the experiment. Seven methods are employed as comparisons. The results are analyzed from both qualitative and quantitative perspectives. Extensive experimental results illustrate that the proposed method performs best on all datasets.

show abstract

“…The ROC curve is the most commonly used evaluation criterion used in the area of hyperspectral anomaly detection [51]. It shows the relationship between the FAR and the true positive rate (TPR) under different threshold.…”

Section: B Evaluation Criteria and Experimental Settingsmentioning

confidence: 99%

A Spectral–Spatial Anomaly Target Detection Method Based on Fractional Fourier Transform and Saliency Weighted Collaborative Representation for Hyperspectral Images

Zhao

Feng

et al. 2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Anomaly target detection methods for hyperspectral images (HSI) often have the problems of potential anomalies and noise contamination when representing background. Therefore, a spectral-spatial hyperspectral anomaly detection method is proposed in this article, which is based on fractional Fourier transform (FrFT) and saliency weighted collaborative representation. First, hyperspectral pixels are projected to the fractional Fourier domain by the FrFT, which can enhance the capability of the detector to suppress the noise and make anomalies to be more distinctive. Then, a saliency weighted matrix is designed as the regularization matrix referring to context-aware saliency theory and combined with the FrFT-based collaborative representation detector. The saliencyweighted regularization matrix assigns different pixels with different weights by using both spectral and spatial information, which can reduce the influence of the potential anomalous pixels embedded in the background when applying collaborative representation theory. Finally, to further improve the performance of the proposed method, a spectral-spatial detection procedure is employed to calculate final anomaly scores by using both spectral information and spatial information. The proposed method is compared with nine state-of-the-art hyperspectral anomaly detection methods on six HSI datasets, including two synthetic HSI datasets and four real-world HSI datasets. Extensive experimental results illustrate that the proposed method's detection performance outperforms other nine well-known compared methods in terms of area under the receiver operating characteristic (ROC) curve values, visual detection characteristics, ROC curve, and separability.

show abstract

Hyperspectral Anomaly Detection with Harmonic Analysis and Low-Rank Decomposition

Cited by 9 publications

References 57 publications

Unsupervised Generative Adversarial Network with Background Enhancement and Irredundant Pooling for Hyperspectral Anomaly Detection

Unsupervised Generative Adversarial Network with Background Enhancement and Irredundant Pooling for Hyperspectral Anomaly Detection

Subfeature Ensemble-Based Hyperspectral Anomaly Detection Algorithm

A Spectral–Spatial Anomaly Target Detection Method Based on Fractional Fourier Transform and Saliency Weighted Collaborative Representation for Hyperspectral Images

Contact Info

Product

Resources

About