Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | Computer Vision |
| Untertitel | ACCV 2014 Workshops |
| Herausgeber (Verlag) | Springer Verlag |
| Seiten | 375-389 |
| Seitenumfang | 15 |
| ISBN (Print) | 9783319166308 |
| Publikationsstatus | Veröffentlicht - 11 Apr. 2015 |
| Veranstaltung | 12th Asian Conference on Computer Vision, ACCV 2014 - Singapore, Singapur Dauer: 1 Nov. 2014 → 5 Nov. 2014 |
Publikationsreihe
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Band | 9009 |
| ISSN (Print) | 0302-9743 |
| ISSN (elektronisch) | 1611-3349 |
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.
ASJC Scopus Sachgebiete
- Mathematik (insg.)
- Theoretische Informatik
- Informatik (insg.)
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
Computer Vision: ACCV 2014 Workshops. Springer Verlag, 2015. S. 375-389 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Band 9009).
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Sparse optimization for motion segmentation
AU - Yang, Michael Ying
AU - Feng, Sitong
AU - Rosenhahn, Bodo
N1 - Funding information: The work is funded by the ERC-Starting Grant (‘DYNAMIC MINVIP’). The authors gratefully acknowledge the support.
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 - http://www.scopus.com/inward/record.url?scp=84942511511&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-16631-5_28
DO - 10.1007/978-3-319-16631-5_28
M3 - Conference contribution
AN - SCOPUS:84942511511
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
PB - Springer Verlag
T2 - 12th Asian Conference on Computer Vision, ACCV 2014
Y2 - 1 November 2014 through 5 November 2014
ER -