Random Collective Representation-Based Detector with Multiple Features for Hyperspectral Images

Fang He,Zhongheng Li,Fei Wang,Weizhong Yu,Haojie Hu

doi:10.3390/rs13040721

Abstract

Collaborative representation-based detector (CRD), as the most representative anomaly detection method, has been widely applied in the field of hyperspectral anomaly detection (HAD). However, the sliding dual window of the original CRD introduces high computational complexity. Moreover, most HAD models only consider a single spectral or spatial feature of the hyperspectral image (HSI), which is unhelpful for improving detection accuracy. To solve these problems, in terms of speed and accuracy, we propose a novel anomaly detection approach, named Random Collective Representation-based Detector with Multiple Feature (RCRDMF). This method includes the following steps. This method first extract the different features include spectral feature, Gabor feature, extended multiattribute profile (EMAP) feature, and extended morphological profile (EMP) feature matrix from the HSI image, which enables us to improve the accuracy of HAD by combining the multiple spectral and spatial features. The ensemble and random collaborative representation detector (ERCRD) method is then applied, which can improve the anomaly detection speed. Finally, an adaptive weight approach is proposed to calculate the weight for each feature. Experimental results on six hyperspectral datasets demonstrate that the proposed approach has the superiority over accuracy and speed.

Highlights

Hyperspectral imagery (HSI), a term that refers to images captured by hyperspectral sensors, can provide rich spectral information for use in identifying different materials with thousands of adjacent contiguous electromagnetic spectrum bands [1,2,3,4,5]
Let X = [ x1, x2, · · ·, xn ] ∈ Rd×n represent a twodimensional HSI matrix that is transformed via three-dimensional hyperspectral imagery, where n and d represent the number of the pixels and spectral bands, respectively
AVIRIS-I: This image was captured from the San Diego airport area, CA, USA, by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor

Summary

Introduction

Hyperspectral imagery (HSI), a term that refers to images captured by hyperspectral sensors, can provide rich spectral information for use in identifying different materials with thousands of adjacent contiguous electromagnetic spectrum bands [1,2,3,4,5]. Depending on whether or not prior information is used, target detection methods can be divided into supervised and unsupervised methods [16]. Supervised target detection methods utilize the known spectral information to detect targets [17], while unsupervised methods detect anomalies from their surrounding background without any prior information [2,18]. It is difficult to obtain the precise spectral information for the supervised method. The GRX, LSMAD, ERCRD, and RCRDMF methods can clearly detect the locations and shapes of the three airplanes. GRX, LSMAD, and ERCRD mistakenly identify many background points as anomalies, while RCRDMF can obtain a clearer detection result. RCRDMF obtains an obvious background-anomaly separation map. RCRDMF performs best in terms of both qualitative and quantitative results

Methods

Results

Discussion

Conclusion