文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Web-based Video Database Management Issues, Mechanisms, and Experimental £ Z Development
侵权投诉

Web-based Video Database Management Issues, Mechanisms, and Experimental £ Z Development

A comprehensive video retrieval system over the web should be able to accommodate and utilize various (complementary) description data in facilitating effective retrieval. In this paper, we advocate a hybrid retrieval approach by integrating a query-based

Web-based Video Database Management: Issues, Mechanisms, and Experimental

£Z

Development

(invited paper)

Qing Li Shermann S.M. ChanDepartment of Computer ScienceCity University of Hong Kong, China

Yi Wu Yueting ZhuangDepartment of Computer ScienceZhejiang University, Hangzhou, China

however actual application demonstrates its importance. At themoment, this aspect of work still remains an unexplored butchallenging problem. Recently, Simone and Ramesh [SJ00] haveattempted to integrate both for image retrieval. They used labelsattached in the HTML files as the textual information, and developedseveral search engines which analyze low-level image featuresindividually. As the labels extracted from the HTML files may not berelevant to the images, the two kinds of information used in theirsystem may not actually represent each other.

In the context of videos, the original research work was on videostructuring, such as shot detection and key frame extraction based onvisual features in the video. Its application areas have been extendedfrom commercial video databases to home videos (e.g., [MZ00]). Butmost of the practical systems annotate videos by textual informationmanually. For example, in OVID[OT93], each video object has a setof manually annotated attributes and attribute values describing itscontent. In [CJ91], both the video objects and their spatialrelationships are manually annotated in order to support complexspatial queries. In order to index video automatically, not only thevisual contents but also the audio content should be used (see [NK98,NC97,WZ01]). Recent work also proposed the use of closed caption.An obvious trend in video indexing and retrieval is to use all thedifferent features like audio, closed captions and so on instead ofsimply visual features. Another trend is towards web-enabled/basedvideo management system.

Over the last couple of years we have been working on developing ageneric video management and application processing (VideoMAP)framework [CL99a,CL99b,LLS00]. A central component ofVideoMAP is a query-based video retrieval mechanism calledCAROL/ST which supports spatio-temporal queries [CL99a,CL99b].While the original CAROL/ST has contributed on working with videosemantic data based on an extended object oriented approach, littlesupport has been provided to support video retrieval using visualfeatures. To support a more effective video retrieval system through ahybrid approach, we have been making extensions to the VideoMAPframework, and particularly the CAROL/ST mechanism [CWLZ01a].In this paper, we further put forward our web-based version ofVideoMAP (termed as VideoMAP*), and discuss the main issuesinvolved in developing such a web-based video database managementsystem supporting hybrid retrieval.

Abstract

A comprehensive video retrieval system over the web should be able toaccommodate and utilize various (complementary) description data infacilitating effective retrieval. In this paper, we advocate a hybridretrieval approach by integrating a query-based (database) mechanismwith content-based retrieval (CBR) functions. We describe the VideoMAP*architecture, discuss issues related to developing such a web-based videodatabase management system, and its specific language mechanism(CAROL/ST with CBR) which provides an improved expressive powerthan what pure query-based or CBR methods currently offer. Anexperimental prototype is being developed using a commercial object-oriented toolkit based on VC++ and Java.

1.Introduction

A current important trend in multimedia information management istowards web-based/enabled multimedia search and managementsystems. Video is a rich and colorful media widely used in many ourdaily life applications like education, entertainment, news spreading,etc. Digital videos have diverse sources of origin such as cassetterecorder, tape recorder, home video camera, VCD and Internet.Expressiveness of video documents decides their dominative positionin the next-generation multimedia information systems. Unliketraditional / static types of data, digital video can provide moreeffective dissemination of information for its rich content.Collectively, a (digital) video can have several informationdescriptors: (1) meta data - the actual video frame stream, includingits encoding scheme and frame rate; (2) media data - the informationabout the characteristic of video content, such as visual feature, scenestructure and spatio-temporal feature; (3) semantic data - the textannotation relevant to the content of video, obtained by manual orautomatic understanding.

Video meta data is created independently from how its contents aredescribed and how its database structure is organized later. It is thusnatural to define “video” and other meaningful constructs such as“scene”, “frame” as objects corresponding to their respective inherentsemantic and visual contents. Meaningful video scenes are identifiedand associated with their description data incrementally. But the gapbetween the user realization and video content remains a big problem.Depending on the user’s viewpoint, the same video/scene may begiven different descriptions. It is extremely difficult (if notimpossible) to describe the whole contents of a video, especially dueto the visual content.

In order to develop an effective (web-based) video retrieval system,one should go beyond the traditional query-based or purely content-based retrieval (CBR) paradigm. Our standpoint is that videos aremulti-faceted data objects, and an effective retrieval system should beable to accommodate all of the above complementary informationdescriptions for retrieving video. As such, in this paper we advocate aweb-based hybrid approach to video retrieval by integrating thequery-based (database) approach with the CBR paradigm.

1.2 Paper Organization

The rest of our paper is organized as follows. In next section wereview the main extensions we have made to VideoMAP, the result ofwhich is a comprehensive video database management system (calledVideoMAP+ ) which supports CBR and query-based retrievalmethods. In section 3 we detail the VideoMAP* framework and itsweb-based hybrid approach to video retrieval. The specific userinterface facilities are described in section 4; example queries are alsogiven to illustrate the expressive power of this mechanism. Finally,we conclude the paper and offer further research directions in section5.

1.1 Related Work

Text-based information retrieval and content-based image retrievalhave been two separate topics historically. The former one usesannotation/specification of high-level semantics of images manually.The latter one needs accurate recognition of low-level image features.Seldom research focuses on these two information integration,

2. Research Background

As mentioned earlier, we have been developing an extended versionof VideoMAP, which we termed it VideoMAP+ [CWLZ01a], for

£Z

This research has been supported, in part, by a Strategic Research Grant from CityU (Project no. 7001073).

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email