Catálogo de publicaciones - libros

The overarching goal of the Informedia Digital Video Library project has been to achieve machine understanding of video media, including all aspects of search, retrieval, visualization and summarization in both contemporaneous and archival content collections. The base technology developed by the Informedia project combines speech, image and natural language understanding to automatically transcribe, segment and index broadcast video for intelligent search and image retrieval. While speech processing has been the most influential component in the success of the Informedia project, other modalities can be critical in various situations. Evaluations done in the context of the TRECVID benchmarks show that while some progress has been made, there is still a lot of work ahead. The fundamental “semantic gap” still exists, but there are a number of promising approaches to bridging it.

Information Retrieval is a supporting technique which underpins a broad range of content-based applications including retrieval, filtering, summarisation, browsing, classification, clustering, automatic linking, and others. Multimedia information retrieval (MMIR) represents those applications when applied to multimedia information such as image, video, music, etc. In this presentation and extended abstract we are primarily concerned with MMIR as applied to information in digital video format. We begin with a brief overview of large scale evaluations of IR tasks in areas such as text, image and music, just to illustrate that this phenomenon is not just restricted to MMIR on video. The main contribution, however, is a set of pointers and a summarisation of the work done as part of TRECVid, the annual benchmarking exercise for video retrieval tasks.

Mobile applications beyond pure mobile telephony are becoming more and more popular for everyday users. In recent years, with the advent of 3G mobile networks such as UMTS and also higher computing power and storage capabilities of mobile devices, multimedia has reached the mobile user. The user’s individual usage context and needs are becoming more and more important for the design of mobile applications. However, the concepts needed to achieve real comprehensiv mobile applications are just evolving. In this paper, we present selected concepts and prototypes from our research in the field of mobile multimedia systems that specifically address the mobile user’s needs. We shortly discuss the aspects of user-centered mobile applications and the challenges we see. Our selected research approaches and prototypes show different concepts towards better supporting the concrete user by mobile applications. In this context, we take a look on the specific challenges of image and video retrieval that arise from placing the user in the center of the mobile application design. In our point of view, user-centered mobile multimedia applications pose interesting challenges not only to the retrieval of multimedia content but introduce new challenges and potentials from acquisition, enhancement, storage, retrieval and delivery to the usage of mobile multimedia content.

The popularity of digital media (images, video, audio) is growing in all segments of the market including consumer, media enterprise, traditional enterprise and Web. Its tremendous growth is a result of the convergence of many factors, including the pervasive increase in bandwidth to users, general affordability of multimedia-ready devices throughout the digital media value chain (creation, management, and distribution), growing ease and affordability of creating digital media content, and growing expectation of the value of digital media in enhancing traditional unstructured and structured information. However, while digital media content is being created and distributed at far greater amounts than ever before, significant technical challenges remain for realizing its full business potential. This paper examines some of the research challenges for industry towards harnessing the full value of digital media.

Just one decade ago image and video retrieval was a technology looking for applications. Now people are dying to get image and video retrieval technology, but there are no good practical solutions. Advances in devices, processing, and storage have resulted in pervasive use of visual information acquisition and usage, but technology development in this area has not kept pace with the rate of other developments. In this paper, we will present some practical systems that are emerging for image and video search and management. I will also present perspectives on why research in image and video retrieval is becoming irrelevant to real world applications. Finally, I will present my beliefs about how research in image and video retrieval can be on the center stage in visual information management for real applications.

Video story segmentation, i.e., segmentation of video to semantically meaningful units, is an essential technology for advanced video processing, such as video retrieval, summarization, and so on. In this paper, we will introduce a generic video story segmentation method, which has achieved highly accurate segmentation on both broadcast news and non-news variety TV programs. Furthermore, we will probe the problems which need to be solved in order to implement story segmentation to practical applications.

To provide easy access to scenes of interest in recorded video, structure-sensitive segmentation is necessary. In TV programs, similar shots appear repeatedly, and such appearance can be a clue to estimate a contextual group of shots. The author introduces a measurement which denotes activeness of shot interaction and enables finding of dialog scenes automatically. This paper presents an algorithm and experimental results of the system which effectively and rapidly detects boundaries of sections in news programs and variety shows.

Many DRM (Digital Rights Management) technologies exist today. While some are consortium standards like DVD-CCA, DTCP, DCP, AACS, others are proprietary like Microsoft’s DRM and Sony’s OMG. They are being used either by different industries or by individual company. Recently, open DRM standards have been developed. OMA DRM has completed its version 2 and many mobile manufacturers are implementing it now, while MPEG IPMP group has completed its MPEG-2 IPMP and MPEG-4 IPMP and is now working on MPEG-21 IPMP. This talk will discuss the design, technology and applications of MPEG IPMP, and how to make DRM successful.

In this talk, we highlight some of the research in I2R on media semantics including sports video analysis, commercial video identification, music retrieval and summarization, image indexing, retrieval and annotation. We then introduce our recent efforts on the international standard called JPSearch which is a new project under ISO/IEC SC29 WG1 (JPEG).

It is now accepted that the most effective video shot retrieval is based on indexing and retrieving clips using multiple, parallel modalities such as text-matching, image-matching and feature matching and then combining or fusing these parallel retrieval streams in some way. In this paper we investigate a range of fusion methods for combining based on multiple visual features (colour, edge and texture), for combining based on multiple visual examples in the query and for combining multiple modalities (text and visual). Using three TRECVid collections and the TRECVid search task, we specifically compare fusion methods based on normalised score and rank that use either the average, weighted average or maximum of retrieval results from a discrete Jelinek-Mercer smoothed language model. We also compare these results with a simple probability-based combination of the language model results that assumes all features and visual examples are fully independent.