Unsupervised Pixel-wise Hyperspectral Anomaly Detection via Autoencoding Adversarial Networks

TL;DR

This paper introduces an unsupervised pixel-wise hyperspectral anomaly detection method using autoencoding adversarial networks across spectral, spatial, and joint domains, outperforming existing detectors.

Contribution

It develops a novel unsupervised hyperspectral anomaly detection approach with three deep autoencoding adversarial models for spectral, spatial, and joint spectral-spatial analysis.

Findings

Outperforms classical and deep learning-based detectors on real datasets

Effective background purification enhances unsupervised training

Reconstruction error maps enable accurate anomaly detection

Abstract

We propose a completely unsupervised pixel-wise anomaly detection method for hyperspectral images. The proposed method consists of three steps called data preparation, reconstruction, and detection. In the data preparation step, we apply a background purification to train the deep network in an unsupervised manner. In the reconstruction step, we propose to use three different deep autoencoding adversarial network (AEAN) models including 1D-AEAN, 2D-AEAN, and 3D-AEAN which are developed for working on spectral, spatial, and joint spectral-spatial domains, respectively. The goal of the AEAN models is to generate synthesized hyperspectral images (HSIs) which are close to real ones. A reconstruction error map (REM) is calculated between the original and the synthesized image pixels. In the detection step, we propose to use a WRX-based detector in which the pixel weights are obtained according to REM. We compare our proposed method with the classical RX, WRX, support vector data description-based (SVDD), collaborative representation-based detector (CRD), adaptive weight deep belief network (AW-DBN) detector and deep autoencoder anomaly detection (DAEAD) method on real hyperspectral datasets. The experimental results show that the proposed approach outperforms other detectors in the benchmark.