Unsupervised and Supervised Feature Extraction Methods for Hyperspectral Images Based on Mixtures of Factor Analyzers

Jocelyn Chanussot,Johannes R Sveinsson,Magnus O Ulfarsson,Bin Zhao

doi:10.3390/rs12071179

Abstract

This paper proposes three feature extraction (FE) methods based on density estimation for hyperspectral images (HSIs). The methods are a mixture of factor analyzers (MFA), deep MFA (DMFA), and supervised MFA (SMFA). The MFA extends the Gaussian mixture model to allow a low-dimensionality representation of the Gaussians. DMFA is a deep version of MFA and consists of a two-layer MFA, i.e, samples from the posterior distribution at the first layer are input to an MFA model at the second layer. SMFA consists of single-layer MFA and exploits labeled information to extract features of HSI effectively. Based on these three FE methods, the paper also proposes a framework that automatically extracts the most important features for classification from an HSI. The overall accuracy of a classifier is used to automatically choose the optimal number of features and hence performs dimensionality reduction (DR) before HSI classification. The performance of MFA, DMFA, and SMFA FE methods are evaluated and compared to five different types of unsupervised and supervised FE methods by using four real HSIs datasets.

Highlights

Hyperspectral images (HSIs) provide abundant spectral information about a scene [1]
We propose an image segmentation method based on the Gaussian mixture model for mixture of factor analyzers (MFA), deep MFA (DMFA), and supervised MFA (SMFA) to solve the problem of a non-normal distribution
Since we were interested in the classification accuracy, we investigated the effect of the number of mixture components M and the dimensionality of latent factors d on overall accuracy (OA)

Summary

Introduction

Hyperspectral images (HSIs) provide abundant spectral information about a scene [1]. In general, an HSI contains hundreds of spectral bands with high spectral resolution [2,3,4]. SMFA is a supervised FE method that uses labeled samples to extract features of HSI Based on these three FE methods, a framework for HSI classification is proposed in this paper. An image segmentation method based on the Gaussian mixture model is proposed for MFA, DMFA, and SMFA to solve the problem of a non-normal distribution. Frameworks for extracting the most useful features for HSI classification based on the MFA, DMFA, and SMFA DR methods are proposed.

Proposed FE Methods And Framework

Framework

Experimental Datasets

Experimental Setup

Tuning Parameter Estimation and Assessment

Classification

Findings

Conclusions