An Outlier-Insensitive Unmixing Algorithm with Spatially Varying Hyperspectral Signatures

Yao Rong Syu, Chia-Hsiang Lin, Chong Yung Chi

Research output: Contribution to journalArticle

Abstract

Effective hyperspectral unmixing (HU) is essential to the estimation of the underlying materials' signatures (endmember signatures) and their spatial distributions (abundance maps) from a given image (data) of a hyperspectral scene. Recently, investigating HU under the non-negligible endmember variability (EV) and outlier effects (OE) has drawn extensive attention. Some state-of-the-art works either consider EV or consider OE, but none of them considers both EV and OE simultaneously. In this paper, we propose a novel HU algorithm, referred to as the variability/outlier-insensitive multi-convex unmixing (VOIMU) algorithm, which is robust against both EV and OE. Considering two suitable regularizers, a nonconvex minimization problem is formulated for which the perturbed linear mixing model proposed by Thouvenin et al., is used for modeling EV, while OE is implicitly handled by applying a p quasi-norm to the data fitting with 0 < p < 1. Then, we reformulate it into a multi-convex problem which is then solved by the block coordinate descent method, with convergence guarantee by casting it into the block successive upper bound minimization framework. The proposed VOIMU algorithm can yield a stationary-point solution with convergence guarantee, together with some intriguing information of potential outlier pixels though outliers are neither physically modeled in the above problem nor detected in the algorithm operation. Finally, we provide some simulation results and experimental results using real data to demonstrate the efficacy and practical applicability of the proposed VOIMU algorithm.

Original languageEnglish
Article number8600733
Pages (from-to)15086-15101
Number of pages16
JournalIEEE Access
Volume7
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

Spatial distribution
Casting
Pixels

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Materials Science(all)
  • Engineering(all)

Cite this

@article{72d611d4bb084ccdbab2acd4d2f45c23,
title = "An Outlier-Insensitive Unmixing Algorithm with Spatially Varying Hyperspectral Signatures",
abstract = "Effective hyperspectral unmixing (HU) is essential to the estimation of the underlying materials' signatures (endmember signatures) and their spatial distributions (abundance maps) from a given image (data) of a hyperspectral scene. Recently, investigating HU under the non-negligible endmember variability (EV) and outlier effects (OE) has drawn extensive attention. Some state-of-the-art works either consider EV or consider OE, but none of them considers both EV and OE simultaneously. In this paper, we propose a novel HU algorithm, referred to as the variability/outlier-insensitive multi-convex unmixing (VOIMU) algorithm, which is robust against both EV and OE. Considering two suitable regularizers, a nonconvex minimization problem is formulated for which the perturbed linear mixing model proposed by Thouvenin et al., is used for modeling EV, while OE is implicitly handled by applying a p quasi-norm to the data fitting with 0 < p < 1. Then, we reformulate it into a multi-convex problem which is then solved by the block coordinate descent method, with convergence guarantee by casting it into the block successive upper bound minimization framework. The proposed VOIMU algorithm can yield a stationary-point solution with convergence guarantee, together with some intriguing information of potential outlier pixels though outliers are neither physically modeled in the above problem nor detected in the algorithm operation. Finally, we provide some simulation results and experimental results using real data to demonstrate the efficacy and practical applicability of the proposed VOIMU algorithm.",
author = "Syu, {Yao Rong} and Chia-Hsiang Lin and Chi, {Chong Yung}",
year = "2019",
month = "1",
day = "1",
doi = "10.1109/ACCESS.2018.2890278",
language = "English",
volume = "7",
pages = "15086--15101",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

An Outlier-Insensitive Unmixing Algorithm with Spatially Varying Hyperspectral Signatures. / Syu, Yao Rong; Lin, Chia-Hsiang; Chi, Chong Yung.

In: IEEE Access, Vol. 7, 8600733, 01.01.2019, p. 15086-15101.

Research output: Contribution to journalArticle

TY - JOUR

T1 - An Outlier-Insensitive Unmixing Algorithm with Spatially Varying Hyperspectral Signatures

AU - Syu, Yao Rong

AU - Lin, Chia-Hsiang

AU - Chi, Chong Yung

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Effective hyperspectral unmixing (HU) is essential to the estimation of the underlying materials' signatures (endmember signatures) and their spatial distributions (abundance maps) from a given image (data) of a hyperspectral scene. Recently, investigating HU under the non-negligible endmember variability (EV) and outlier effects (OE) has drawn extensive attention. Some state-of-the-art works either consider EV or consider OE, but none of them considers both EV and OE simultaneously. In this paper, we propose a novel HU algorithm, referred to as the variability/outlier-insensitive multi-convex unmixing (VOIMU) algorithm, which is robust against both EV and OE. Considering two suitable regularizers, a nonconvex minimization problem is formulated for which the perturbed linear mixing model proposed by Thouvenin et al., is used for modeling EV, while OE is implicitly handled by applying a p quasi-norm to the data fitting with 0 < p < 1. Then, we reformulate it into a multi-convex problem which is then solved by the block coordinate descent method, with convergence guarantee by casting it into the block successive upper bound minimization framework. The proposed VOIMU algorithm can yield a stationary-point solution with convergence guarantee, together with some intriguing information of potential outlier pixels though outliers are neither physically modeled in the above problem nor detected in the algorithm operation. Finally, we provide some simulation results and experimental results using real data to demonstrate the efficacy and practical applicability of the proposed VOIMU algorithm.

AB - Effective hyperspectral unmixing (HU) is essential to the estimation of the underlying materials' signatures (endmember signatures) and their spatial distributions (abundance maps) from a given image (data) of a hyperspectral scene. Recently, investigating HU under the non-negligible endmember variability (EV) and outlier effects (OE) has drawn extensive attention. Some state-of-the-art works either consider EV or consider OE, but none of them considers both EV and OE simultaneously. In this paper, we propose a novel HU algorithm, referred to as the variability/outlier-insensitive multi-convex unmixing (VOIMU) algorithm, which is robust against both EV and OE. Considering two suitable regularizers, a nonconvex minimization problem is formulated for which the perturbed linear mixing model proposed by Thouvenin et al., is used for modeling EV, while OE is implicitly handled by applying a p quasi-norm to the data fitting with 0 < p < 1. Then, we reformulate it into a multi-convex problem which is then solved by the block coordinate descent method, with convergence guarantee by casting it into the block successive upper bound minimization framework. The proposed VOIMU algorithm can yield a stationary-point solution with convergence guarantee, together with some intriguing information of potential outlier pixels though outliers are neither physically modeled in the above problem nor detected in the algorithm operation. Finally, we provide some simulation results and experimental results using real data to demonstrate the efficacy and practical applicability of the proposed VOIMU algorithm.

UR - http://www.scopus.com/inward/record.url?scp=85061785051&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85061785051&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2018.2890278

DO - 10.1109/ACCESS.2018.2890278

M3 - Article

VL - 7

SP - 15086

EP - 15101

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

M1 - 8600733

ER -