An improved EMD-based dissimilarity Metric for Unsupervised Linear Subspace Learning

Xiangchun Yu, Zhezhou Yu, Wei Pang, Minghao Li, Lei Wu

Research output: Contribution to journalArticle

2 Citations (Scopus)
8 Downloads (Pure)

Abstract

We investigate a novel way of robust face image feature extraction by adopting the methods based on Unsupervised Linear Subspace Learning to extract a small number of good features. Firstly, the face image is divided into blocks with the specified size, and then we propose and extract pooled Histogram of Oriented Gradient (pHOG) over each block. Secondly, an improved Earth Mover’s Distance (EMD) metric is adopted to measure the dissimilarity between blocks of one face image and the corresponding blocks from the rest of face images. Thirdly, considering the limitations of the original Locality Preserving Projections (LPP), we proposed the Block Structure LPP (BSLPP), which effectively preserves the structural information of face images. Finally, an adjacency graph is constructed and a small number of good features of a face image are obtained by methods based on Unsupervised Linear Subspace Learning. A series of experiments have been conducted on several well-known face databases to evaluate the effectiveness of the proposed algorithm. In addition, we construct the noise, geometric distortion, slight translation, slight rotation AR, and Extended Yale B face databases, and we verify the robustness of the proposed algorithm when faced with a certain degree of these disturbances.
Original languageEnglish
Article number8917393
Number of pages24
JournalComplexity
Volume2018
DOIs
Publication statusPublished - 18 Feb 2018

Fingerprint

Earth (planet)
Feature extraction
Experiments

Cite this

An improved EMD-based dissimilarity Metric for Unsupervised Linear Subspace Learning. / Yu, Xiangchun; Yu, Zhezhou; Pang, Wei; Li, Minghao; Wu, Lei.

In: Complexity, Vol. 2018, 8917393, 18.02.2018.

Research output: Contribution to journalArticle

Yu, Xiangchun ; Yu, Zhezhou ; Pang, Wei ; Li, Minghao ; Wu, Lei. / An improved EMD-based dissimilarity Metric for Unsupervised Linear Subspace Learning. In: Complexity. 2018 ; Vol. 2018.
@article{7a1af5a28a3941a7bca95bc9eab20b4e,
title = "An improved EMD-based dissimilarity Metric for Unsupervised Linear Subspace Learning",
abstract = "We investigate a novel way of robust face image feature extraction by adopting the methods based on Unsupervised Linear Subspace Learning to extract a small number of good features. Firstly, the face image is divided into blocks with the specified size, and then we propose and extract pooled Histogram of Oriented Gradient (pHOG) over each block. Secondly, an improved Earth Mover’s Distance (EMD) metric is adopted to measure the dissimilarity between blocks of one face image and the corresponding blocks from the rest of face images. Thirdly, considering the limitations of the original Locality Preserving Projections (LPP), we proposed the Block Structure LPP (BSLPP), which effectively preserves the structural information of face images. Finally, an adjacency graph is constructed and a small number of good features of a face image are obtained by methods based on Unsupervised Linear Subspace Learning. A series of experiments have been conducted on several well-known face databases to evaluate the effectiveness of the proposed algorithm. In addition, we construct the noise, geometric distortion, slight translation, slight rotation AR, and Extended Yale B face databases, and we verify the robustness of the proposed algorithm when faced with a certain degree of these disturbances.",
author = "Xiangchun Yu and Zhezhou Yu and Wei Pang and Minghao Li and Lei Wu",
year = "2018",
month = "2",
day = "18",
doi = "10.1155/2018/8917393",
language = "English",
volume = "2018",
journal = "Complexity",
issn = "1076-2787",
publisher = "John Wiley and Sons Inc.",

}

TY - JOUR

T1 - An improved EMD-based dissimilarity Metric for Unsupervised Linear Subspace Learning

AU - Yu, Xiangchun

AU - Yu, Zhezhou

AU - Pang, Wei

AU - Li, Minghao

AU - Wu, Lei

PY - 2018/2/18

Y1 - 2018/2/18

N2 - We investigate a novel way of robust face image feature extraction by adopting the methods based on Unsupervised Linear Subspace Learning to extract a small number of good features. Firstly, the face image is divided into blocks with the specified size, and then we propose and extract pooled Histogram of Oriented Gradient (pHOG) over each block. Secondly, an improved Earth Mover’s Distance (EMD) metric is adopted to measure the dissimilarity between blocks of one face image and the corresponding blocks from the rest of face images. Thirdly, considering the limitations of the original Locality Preserving Projections (LPP), we proposed the Block Structure LPP (BSLPP), which effectively preserves the structural information of face images. Finally, an adjacency graph is constructed and a small number of good features of a face image are obtained by methods based on Unsupervised Linear Subspace Learning. A series of experiments have been conducted on several well-known face databases to evaluate the effectiveness of the proposed algorithm. In addition, we construct the noise, geometric distortion, slight translation, slight rotation AR, and Extended Yale B face databases, and we verify the robustness of the proposed algorithm when faced with a certain degree of these disturbances.

AB - We investigate a novel way of robust face image feature extraction by adopting the methods based on Unsupervised Linear Subspace Learning to extract a small number of good features. Firstly, the face image is divided into blocks with the specified size, and then we propose and extract pooled Histogram of Oriented Gradient (pHOG) over each block. Secondly, an improved Earth Mover’s Distance (EMD) metric is adopted to measure the dissimilarity between blocks of one face image and the corresponding blocks from the rest of face images. Thirdly, considering the limitations of the original Locality Preserving Projections (LPP), we proposed the Block Structure LPP (BSLPP), which effectively preserves the structural information of face images. Finally, an adjacency graph is constructed and a small number of good features of a face image are obtained by methods based on Unsupervised Linear Subspace Learning. A series of experiments have been conducted on several well-known face databases to evaluate the effectiveness of the proposed algorithm. In addition, we construct the noise, geometric distortion, slight translation, slight rotation AR, and Extended Yale B face databases, and we verify the robustness of the proposed algorithm when faced with a certain degree of these disturbances.

U2 - 10.1155/2018/8917393

DO - 10.1155/2018/8917393

M3 - Article

VL - 2018

JO - Complexity

JF - Complexity

SN - 1076-2787

M1 - 8917393

ER -