A United Framework for Video Browsing and Retrieval

Retrieval Yong Rui and Thomas S. Huang Be kman Institute for Advan ed S ien e and Te hnology University of Illinois at Urbana-Champaign Urbana, IL 61801, USA E-mail:fyrui, huangg ifp.uiu .edu 1 Introdu tion Resear h on how to eÆ iently a ess video ontent has be ome in reasingly a tive in the past few years [1, 2, 3, 4℄. Considerable progress has been made in video analysis, representation, browsing, and retrieval, whi h are the four fundamental bases for a essing video ontent. Video analysis deals with the signal pro essing part of the video system, in luding shot boundary dete tion, key frame extra tion, et . Video representation is on erned with the stru ture of the video. An example of a video representation is the tree stru tured key frame hierar hy [5, 3℄. Built on top of the video representation, video browsing deals with how to use the representation stru ture to help viewers browse the video ontent. Finally, video retrieval is on erned with retrieving interesting video obje ts. The relationship between these four resear h areas is illustrated in Figure 1. Video Browsing Video Retrieval