S. Panchanathan
Arizona State University

A Position Statement for Panel 2: Video representation, coding, indexing
The 1998 International Workshop on Very Low Bitrate Video Coding


The increasing usage of visual media in various multimedia applications is challenging researchers to investigate novel methods of representing, storing, indexing and retrieving visual information. Until recently the focus has been largely centered around information theoretic approaches for compression. If representation, indexing and coding all rely on efficient information extraction, shouldn't all of them be tackled jointly instead of optimizing these processes individually ? In other words, if one is able to extract information for indexing, should this not serve as a valuable cue for efficient compression or vice versa? In the context of multimedia data, there is quite a bit to be gained in employing multiple media indices for efficient coding/indexing and retrieval. Should we not exploit inter-media relationships for enhanced performance rather than limiting ourselves to visual information? Current query, indexing and retrieval systems are inadequate in capturing the needs of the user as well as the user profiles resulting in cumbersome search and/or irrelevant/ inaccurate retrieval. If human is the ultimate client, should we not be focussing on designing systems which abstract human behavior in terms of user search styles, semantics, patterns and preferences rather than developing one other sophisticated compression algorithm which provides a 0.5 dB improvement in performance or a marginally better indexing/retrieval algorithm ?