Abstract

In this paper, we propose a new anomaly detection method for hyperspectral images based on two well-designed dictionaries: background dictionary and potential anomaly dictionary. In order to effectively detect an anomaly and eliminate the influence of noise, the original image is decomposed into three components: background, anomalies, and noise. In this way, the anomaly detection task is regarded as a problem of matrix decomposition. Considering the homogeneity of background and the sparsity of anomalies, the low-rank and sparse constraints are imposed in our model. Then, the background and potential anomaly dictionaries are constructed using the background and anomaly priors. For the background dictionary, a joint sparse representation (JSR)-based dictionary selection strategy is proposed, assuming that the frequently used atoms in the overcomplete dictionary tend to be the background. In order to make full use of the prior information of anomalies hidden in the scene, the potential anomaly dictionary is constructed. We define a criterion, i.e., the anomalous level of a pixel, by using the residual calculated in the JSR model within its local region. Then, it is combined with a weighted term to alleviate the influence of noise and background. Experiments show that our proposed anomaly detection method based on potential anomaly and background dictionaries construction can achieve superior results compared with other state-of-the-art methods.

Highlights

  • H YPERSPECTRAL images (HSIs) are of wide spectral range and high spectral resolution [1][2], so it contains rich spectral information to discriminate physical properties of different materials [3]

  • To accurately model background and anomaly information, we propose a new HSI anomaly detection method based on well-designed background and potential anomaly dictionaries utilizing low rank and sparse representation strategy

  • 3) Making full use of the anomaly prior information hidden in the scene, we propose to construct a potential anomaly dictionaries utilizing the residual of each local region by the joint sparse representation (JSR) model

Read more

Summary

INTRODUCTION

H YPERSPECTRAL images (HSIs) are of wide spectral range and high spectral resolution [1][2], so it contains rich spectral information to discriminate physical properties of different materials [3]. For HSI data, an LRR technique has been used in classification [40][41] and denoising [42] This method was utilized to model the problem of anomaly detection [43][44] based on the assumption that the background has low rank properties whilst the anomalies demonstrate sparse properties. To accurately model background and anomaly information, we propose a new HSI anomaly detection method based on well-designed background and potential anomaly dictionaries utilizing low rank and sparse representation strategy. 1) A new low rank and sparsity based anomaly detection model is proposed with two well designed dictionaries, i.e., background and potential anomaly dictionaries, so that the original data can be properly decomposed into background, anomaly, and noise components.

PROPOSED METHOD
Background dictionary construction
Potential anomaly dictionary construction
Optimization and computational complexity
EXPERIMENTS RESULTS
Dataset description
Detection performance
Parameter Analysis
Wi7ndow9size
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call