Advances in Visual Information Management: Visual Database by Masao Sakauchi (auth.), Hiroshi Arisawa, Tiziana Catarci

Posted by

By Masao Sakauchi (auth.), Hiroshi Arisawa, Tiziana Catarci (eds.)

Video segmentation is the main basic method for applicable index­ ing and retrieval of video durations. in most cases, video streams are composed 1 of pictures delimited by way of actual shot barriers. titanic paintings has been performed on easy methods to observe such shot barriers immediately (Arman et aI. , 1993) (Zhang et aI. , 1993) (Zhang et aI. , 1995) (Kobla et aI. , 1997). in the course of the inte­ gration of applied sciences resembling photograph processing, speech/character reputation and common language figuring out, key words might be extracted and linked to those photographs for indexing (Wactlar et aI. , 1996). A unmarried shot, despite the fact that, hardly consists of sufficient volume of data to be significant on its own. Usu­ best friend, it's a semantically significant period that almost all clients have an interest in re­ trieving. as a rule, such significant durations span numerous consecutive pictures. There infrequently exists any effective and trustworthy approach, both computerized or guide, to spot all semantically significant periods inside of a video movement. Works by means of (Smith and Davenport, 1992) (Oomoto and Tanaka, 1993) (Weiss et aI. , 1995) (Hjelsvold et aI. , 1996) recommend manually defining all such inter­ vals within the database prematurely. even though, even an hour lengthy video could have an indefinite variety of significant periods. furthermore, video info is multi­ interpretative. for this reason, given a question, what's a significant period to an annotator will not be significant to the consumer who matters the question. In perform, handbook indexing of significant periods is labour extensive and inadequate.

Show description

Read Online or Download Advances in Visual Information Management: Visual Database Systems. IFIP TC2 WG2.6 Fifth Working Conference on Visual Database Systems May 10–12, 2000, Fukuoka, Japan PDF

Similar japan books

Inside the Robot Kingdom: Japan, Mechatronics, and the Coming Robotopia

It is a electronic re-issue of Frederik L. Schodt's now-classic paintings, "Inside the robotic country: Japan Mechatronics, and the arriving Robotopia. " concentrating on business and fable robots in Japan as a metaphor for the connection among tradition and know-how, this publication used to be first issued in hardcover in 1988, and paperback in 1990.

Advances in Visual Information Management: Visual Database Systems. IFIP TC2 WG2.6 Fifth Working Conference on Visual Database Systems May 10–12, 2000, Fukuoka, Japan

Video segmentation is the main basic technique for applicable index­ ing and retrieval of video periods. usually, video streams are composed 1 of pictures delimited by means of actual shot limitations. enormous paintings has been performed on easy methods to notice such shot obstacles immediately (Arman et aI. , 1993) (Zhang et aI.

Japan ist offen: Chancen für deutsche Unternehmen

Seit zwei Jahrzehnten ist Japan die zweitgrößte Volkswirtschaft der Welt. Japan erarbeitet 70% der Wirtschaftsleistung der im Aufbruch befindlichen ost- und südostasiatischen quarter. Wiederum 60% des japanischen Inlandsprodukts werden für den privaten Verbrauch verwendet. Dies bedeutet ein erhebliches Marktpotential auch für die deutsche Wirtschaft.

Additional resources for Advances in Visual Information Management: Visual Database Systems. IFIP TC2 WG2.6 Fifth Working Conference on Visual Database Systems May 10–12, 2000, Fukuoka, Japan

Example text

1996) A Survey of Technologies for Parsing and Indexing Digital Video; Journal of Visual Communication and Image Representation, (Special Issue on Digital Libraries), March 1996, Vol. 1, pp. , Ono, K. (1997) Phasme: Un Systeme Parallele de Gestion de Bases de Donnees Oriente Application, Calculateurs Paralleles, Special Issue on Parallel and Distributed Database Systems, 1997. Andres, F. , Ono, K. (1998) Active Hypermedia Delivery System; Proc. of the Int'l Conf. On Data Engineering (ICDE'98), Florida, February 1998, pp.

2 Content-based representation Storage The content-based representation includes three sets of data: the face information (FaceMinPosition, FaceMaxPosition, and Frame), the Face name information (Name) and the relationship with Video Shots. 'All those informations are stored inside the Extended Binary Graph (EBG) data structure (Andres and Ono 97) to provide high performance access for the retrieval process. g. Name, Video shots); and (3) Access from IDs to values and reverse access from values to ID.

In this paper, we use motion vectors including an MPEG encoded video stream as optical flow to obtain camera parameters. 3. Similar Scene Detection Scenes similar to the retrieval key are detected by feature matching between the extracted two series of camera parameters. Though several methods have been proposed to perform matching between two series of features, we adopt the continuous DP matching method since it effectively detects similar scenes by taking both motion transition similarity and differences in scene length into consideration.

Download PDF sample

Rated 4.58 of 5 – based on 50 votes