Sparse optimization for motion segmentation

Michael Ying Yang, Sitong Feng, Bodo Rosenhahn

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

1 Citation (Scopus)

Abstract

In this paper, we propose a new framework for segmenting feature-based multiple moving objects with subspace models in affine views. Since the feature data is high-dimensional and complex in the real video sequences, most traditional approaches for motion segmentation use the conventional PCA to obtain a low-dimensional representation, while our proposed framework applies the sparse PCA (SPCA) to obtain a projected subspace, which is a low-dimensional global subspace on a Stiefel manifold with sparse entries. Then, the local subspace separation is achieved via automatically selecting the sparse nearest neighbours. By combining two sparse techniques, the proposed framework segments different motions through a simple spectral clustering on an affinity matrix built with the principal angles. To the best of our knowledge, our framework is the first one to apply the sparse optimization for optimizing the global and local subspace simultaneously.We test our method extensively and compare its performance to several state-of-art motion segmentation methods with experiments on the Hopkins 155 dataset. Our results are comparable with these results, and in many cases exceed them both in terms of segmentation accuracy and computational speed.

Original languageEnglish
Title of host publicationComputer Vision - ACCV 2014 Workshops
Subtitle of host publicationRevised Selected Papers, Part II
EditorsC.V. Jawahar, Shiguang Shan
Place of PublicationSingapore
PublisherSpringer Verlag
Pages375-389
Number of pages15
ISBN (Print)9783319166308
DOIs
Publication statusPublished - 11 Apr 2015
Externally publishedYes
Event12th Asian Conference on Computer Vision, ACCV 2014 - Singapore, Singapore
Duration: 1 Nov 20145 Nov 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9009
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference12th Asian Conference on Computer Vision, ACCV 2014
Abbreviated titleACCV 2014
CountrySingapore
CitySingapore
Period1/11/145/11/14

Fingerprint

Motion Segmentation
Subspace
Optimization
Experiments
Stiefel Manifold
Spectral Clustering
Moving Objects
Affine transformation
Nearest Neighbor
Exceed
High-dimensional
Segmentation
Angle
Motion
Framework
Experiment

Cite this

Yang, M. Y., Feng, S., & Rosenhahn, B. (2015). Sparse optimization for motion segmentation. In C. V. Jawahar, & S. Shan (Eds.), Computer Vision - ACCV 2014 Workshops: Revised Selected Papers, Part II (pp. 375-389). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9009). Singapore: Springer Verlag. https://doi.org/10.1007/978-3-319-16631-5_28
Yang, Michael Ying ; Feng, Sitong ; Rosenhahn, Bodo. / Sparse optimization for motion segmentation. Computer Vision - ACCV 2014 Workshops: Revised Selected Papers, Part II. editor / C.V. Jawahar ; Shiguang Shan. Singapore : Springer Verlag, 2015. pp. 375-389 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{e4c70c70b69c4345b720dc016238e0ba,
title = "Sparse optimization for motion segmentation",
abstract = "In this paper, we propose a new framework for segmenting feature-based multiple moving objects with subspace models in affine views. Since the feature data is high-dimensional and complex in the real video sequences, most traditional approaches for motion segmentation use the conventional PCA to obtain a low-dimensional representation, while our proposed framework applies the sparse PCA (SPCA) to obtain a projected subspace, which is a low-dimensional global subspace on a Stiefel manifold with sparse entries. Then, the local subspace separation is achieved via automatically selecting the sparse nearest neighbours. By combining two sparse techniques, the proposed framework segments different motions through a simple spectral clustering on an affinity matrix built with the principal angles. To the best of our knowledge, our framework is the first one to apply the sparse optimization for optimizing the global and local subspace simultaneously.We test our method extensively and compare its performance to several state-of-art motion segmentation methods with experiments on the Hopkins 155 dataset. Our results are comparable with these results, and in many cases exceed them both in terms of segmentation accuracy and computational speed.",
author = "Yang, {Michael Ying} and Sitong Feng and Bodo Rosenhahn",
year = "2015",
month = "4",
day = "11",
doi = "10.1007/978-3-319-16631-5_28",
language = "English",
isbn = "9783319166308",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "375--389",
editor = "C.V. Jawahar and Shiguang Shan",
booktitle = "Computer Vision - ACCV 2014 Workshops",
address = "Germany",

}

Yang, MY, Feng, S & Rosenhahn, B 2015, Sparse optimization for motion segmentation. in CV Jawahar & S Shan (eds), Computer Vision - ACCV 2014 Workshops: Revised Selected Papers, Part II. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9009, Springer Verlag, Singapore, pp. 375-389, 12th Asian Conference on Computer Vision, ACCV 2014, Singapore, Singapore, 1/11/14. https://doi.org/10.1007/978-3-319-16631-5_28

Sparse optimization for motion segmentation. / Yang, Michael Ying; Feng, Sitong; Rosenhahn, Bodo.

Computer Vision - ACCV 2014 Workshops: Revised Selected Papers, Part II. ed. / C.V. Jawahar; Shiguang Shan. Singapore : Springer Verlag, 2015. p. 375-389 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9009).

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Sparse optimization for motion segmentation

AU - Yang, Michael Ying

AU - Feng, Sitong

AU - Rosenhahn, Bodo

PY - 2015/4/11

Y1 - 2015/4/11

N2 - In this paper, we propose a new framework for segmenting feature-based multiple moving objects with subspace models in affine views. Since the feature data is high-dimensional and complex in the real video sequences, most traditional approaches for motion segmentation use the conventional PCA to obtain a low-dimensional representation, while our proposed framework applies the sparse PCA (SPCA) to obtain a projected subspace, which is a low-dimensional global subspace on a Stiefel manifold with sparse entries. Then, the local subspace separation is achieved via automatically selecting the sparse nearest neighbours. By combining two sparse techniques, the proposed framework segments different motions through a simple spectral clustering on an affinity matrix built with the principal angles. To the best of our knowledge, our framework is the first one to apply the sparse optimization for optimizing the global and local subspace simultaneously.We test our method extensively and compare its performance to several state-of-art motion segmentation methods with experiments on the Hopkins 155 dataset. Our results are comparable with these results, and in many cases exceed them both in terms of segmentation accuracy and computational speed.

AB - In this paper, we propose a new framework for segmenting feature-based multiple moving objects with subspace models in affine views. Since the feature data is high-dimensional and complex in the real video sequences, most traditional approaches for motion segmentation use the conventional PCA to obtain a low-dimensional representation, while our proposed framework applies the sparse PCA (SPCA) to obtain a projected subspace, which is a low-dimensional global subspace on a Stiefel manifold with sparse entries. Then, the local subspace separation is achieved via automatically selecting the sparse nearest neighbours. By combining two sparse techniques, the proposed framework segments different motions through a simple spectral clustering on an affinity matrix built with the principal angles. To the best of our knowledge, our framework is the first one to apply the sparse optimization for optimizing the global and local subspace simultaneously.We test our method extensively and compare its performance to several state-of-art motion segmentation methods with experiments on the Hopkins 155 dataset. Our results are comparable with these results, and in many cases exceed them both in terms of segmentation accuracy and computational speed.

UR - https://ezproxy2.utwente.nl/login?url=https://doi.org/10.1007/978-3-319-16631-5_28

UR - https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2015/chap/yang_spa.pdf

U2 - 10.1007/978-3-319-16631-5_28

DO - 10.1007/978-3-319-16631-5_28

M3 - Conference contribution

SN - 9783319166308

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 375

EP - 389

BT - Computer Vision - ACCV 2014 Workshops

A2 - Jawahar, C.V.

A2 - Shan, Shiguang

PB - Springer Verlag

CY - Singapore

ER -

Yang MY, Feng S, Rosenhahn B. Sparse optimization for motion segmentation. In Jawahar CV, Shan S, editors, Computer Vision - ACCV 2014 Workshops: Revised Selected Papers, Part II. Singapore: Springer Verlag. 2015. p. 375-389. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16631-5_28