Video saliency detection with robust temporal alignment and local-global spatial contrast
Video saliency detection, the task to detect attractive content in a video, has broad applications in multimedia understanding and retrieval. In this paper, we propose a new framework for spatiotemporal saliency detection. To better estimate the salient motion in temporal domain, we take advantage o...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Language: | English |
| Published: |
2013
|
| Online Access: | https://hdl.handle.net/10356/98776 http://hdl.handle.net/10220/12673 |
| _version_ | 1826121815370498048 |
|---|---|
| author | Ren, Zhixiang Chia, Clement Liang-Tien Rajan, Deepu |
| author2 | School of Computer Engineering |
| author_facet | School of Computer Engineering Ren, Zhixiang Chia, Clement Liang-Tien Rajan, Deepu |
| author_sort | Ren, Zhixiang |
| collection | NTU |
| description | Video saliency detection, the task to detect attractive content in a video, has broad applications in multimedia understanding and retrieval. In this paper, we propose a new framework for spatiotemporal saliency detection. To better estimate the salient motion in temporal domain, we take advantage of robust alignment by sparse and low-rank decomposition to jointly estimate the salient foreground motion and the camera motion. Consecutive frames are transformed and aligned, and then decomposed to a low-rank matrix representing the background and a sparse matrix indicating the objects with salient motion. In the spatial domain, we address several problems of local center-surround contrast based model, and demonstrate how to utilize global information and prior knowledge to improve spatial saliency detection. Individual component evaluation demonstrates the effectiveness of our temporal and spatial methods. Final experimental results show that the combination of our spatial and temporal saliency maps achieve the best overall performance compared to several state-of-the-art methods. |
| first_indexed | 2024-10-01T05:38:20Z |
| format | Conference Paper |
| id | ntu-10356/98776 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:38:20Z |
| publishDate | 2013 |
| record_format | dspace |
| spelling | ntu-10356/987762020-05-28T07:41:41Z Video saliency detection with robust temporal alignment and local-global spatial contrast Ren, Zhixiang Chia, Clement Liang-Tien Rajan, Deepu School of Computer Engineering International Conference on Multimedia Retrieval (2nd : 2012) Video saliency detection, the task to detect attractive content in a video, has broad applications in multimedia understanding and retrieval. In this paper, we propose a new framework for spatiotemporal saliency detection. To better estimate the salient motion in temporal domain, we take advantage of robust alignment by sparse and low-rank decomposition to jointly estimate the salient foreground motion and the camera motion. Consecutive frames are transformed and aligned, and then decomposed to a low-rank matrix representing the background and a sparse matrix indicating the objects with salient motion. In the spatial domain, we address several problems of local center-surround contrast based model, and demonstrate how to utilize global information and prior knowledge to improve spatial saliency detection. Individual component evaluation demonstrates the effectiveness of our temporal and spatial methods. Final experimental results show that the combination of our spatial and temporal saliency maps achieve the best overall performance compared to several state-of-the-art methods. 2013-07-31T09:00:09Z 2019-12-06T19:59:33Z 2013-07-31T09:00:09Z 2019-12-06T19:59:33Z 2012 2012 Conference Paper Ren, Z., Chia, C. L. T., & Rajan, D. (2012). Video saliency detection with robust temporal alignment and local-global spatial contrast. Proceedings of the 2nd ACM International Conference on Multimedia Retrieval - ICMR '12. https://hdl.handle.net/10356/98776 http://hdl.handle.net/10220/12673 10.1145/2324796.2324851 en |
| spellingShingle | Ren, Zhixiang Chia, Clement Liang-Tien Rajan, Deepu Video saliency detection with robust temporal alignment and local-global spatial contrast |
| title | Video saliency detection with robust temporal alignment and local-global spatial contrast |
| title_full | Video saliency detection with robust temporal alignment and local-global spatial contrast |
| title_fullStr | Video saliency detection with robust temporal alignment and local-global spatial contrast |
| title_full_unstemmed | Video saliency detection with robust temporal alignment and local-global spatial contrast |
| title_short | Video saliency detection with robust temporal alignment and local-global spatial contrast |
| title_sort | video saliency detection with robust temporal alignment and local global spatial contrast |
| url | https://hdl.handle.net/10356/98776 http://hdl.handle.net/10220/12673 |
| work_keys_str_mv | AT renzhixiang videosaliencydetectionwithrobusttemporalalignmentandlocalglobalspatialcontrast AT chiaclementliangtien videosaliencydetectionwithrobusttemporalalignmentandlocalglobalspatialcontrast AT rajandeepu videosaliencydetectionwithrobusttemporalalignmentandlocalglobalspatialcontrast |