Graph-based multi-space semantic correlation propagation for video retrieval

doi:10.1007/s00371-010-0510-6

	Graph-based multi-space semantic correlation propagation for video retrieval
	Feng, Bailan 1,2; Cao, Juan 1; Bao, Xiuguo 3; Bao, Lei 1,2; Zhang, Yongdong 1; Lin, Shouxun 1; Yun, Xiaochun 3
	2011
发表期刊	VISUAL COMPUTER
ISSN	0178-2789
卷号	27 期号:1 页码:21-34
摘要	By introducing the concept detection results to the retrieval process, concept-based video retrieval (CBVR) has been successfully used for semantic content-based video retrieval application. However, how to select and fuse the appropriate concepts for a specific query is still an important but difficult issue. In this paper, we propose a novel and effective concept selection method, named graph-based multi-space semantic correlation propagation (GMSSCP), to explore the relationship between the user query and concepts for video retrieval application. Compared with traditional methods, GMSSCP makes use of a manifold-ranking algorithm to collectively explore the multi-layered relationships between the query and concepts, and the expansion result is more robust to noises. Parallel to this, GMSSCP has a query-adapting property, which can enhance the process of concept correlation propagation and selection with strong pertinence of query cues. Furthermore, it can dynamically update the unified propagation graph by flexibly introducing the multi-modal query cues as additional nodes, and is not only effective for automatic retrieval but also appropriate for the interactive case. Encouraging experimental results on TRECVID datasets demonstrate the effectiveness of GMSSCP over the state-of-the-art concept selection methods. Moreover, we also apply it to the interactive retrieval system-VideoMap and gain an excellent performance and user experience.
关键词	Concept-based video retrieval Concept selection and fusion Multi-space integration and propagation Manifold ranking
DOI	10.1007/s00371-010-0510-6
收录类别	SCI
语种	英语
资助项目	National Basic Research Program of China (973 Program)[2007CB311100] ; National High Technology and Research Development Program of China (863 Program)[2007AA01Z416] ; National Nature Science Foundation of China[6087165] ; National Nature Science Foundation of China[60802028] ; National Nature Science Foundation of China[60902090] ; Beijing New Star Project on Science Technology[2007B071] ; Beijing Municipal Education Commission
WOS研究方向	Computer Science
WOS类目	Computer Science, Software Engineering
WOS记录号	WOS:000285781100002
出版者	SPRINGER
引用统计	被引频次：11[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/12865
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Feng, Bailan
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Lab Adv Comp Res, Beijing 100190, Peoples R China 2.Chinese Acad Sci, Grad Univ, Beijing 100039, Peoples R China 3.Coordinat Ctr China, Natl Comp Network Emergency Response Tech, Beijing 100029, Peoples R China
推荐引用方式 GB/T 7714	Feng, Bailan,Cao, Juan,Bao, Xiuguo,et al. Graph-based multi-space semantic correlation propagation for video retrieval[J]. VISUAL COMPUTER,2011,27(1):21-34.
APA	Feng, Bailan.,Cao, Juan.,Bao, Xiuguo.,Bao, Lei.,Zhang, Yongdong.,...&Yun, Xiaochun.(2011).Graph-based multi-space semantic correlation propagation for video retrieval.VISUAL COMPUTER,27(1),21-34.
MLA	Feng, Bailan,et al."Graph-based multi-space semantic correlation propagation for video retrieval".VISUAL COMPUTER 27.1(2011):21-34.