Multimedia Clip Generation from Documents for Browsing on Mobile Devices

Berna Erol, Kathrin Berkner
Ricoh Innovations, California Research Center
2282 Sand Hill Rd., Menlo Park, CA, USA
berna_erol@rii.ricoh.com  berkner@rii.ricoh.com

Siddharth Joshi
Department of Electrical Engineering
Stanford University
Packard 243, 350 Serra Mall, Stanford, CA, USA
sidj@stanford.edu

Abstract

Small displays on mobile handheld devices, such as PDAs and cellular phones, are the bottlenecks for usability of most content browsing applications. Generally, conventional content such as documents and web pages need to be modified for effective presentation on mobile devices. This paper proposes a novel visualization for documents, called Multimedia Thumbnails, which consists of text and image content converted into playable multimedia clips. A Multimedia Thumbnail utilizes visual and audio channels of small portable devices as well as both spatial and time dimensions to communicate text and image information of a single document. The proposed algorithm for generating Multimedia Thumbnails includes 1) a semantic document analysis step, where salient content from a source document is extracted, 2) an optimization step, where a subset of this extracted content is selected based on time, display, and application constraints, and 3) a composition step, where the selected visual and audible document content is combined into a Multimedia Thumbnail. Scalability of MMNails that allows generation of multimedia clips of various lengths is also described. A user study is presented that evaluates the effectiveness of the proposed Multimedia Thumbnail visualization.

EDICS Categories

3-CONT, 4-ANSY , 3-INTF

Index Terms

mobile document, document conversion, adaptive content delivery, document repurposing, multimedia generation, content navigation, multimedia thumbnail.