Jean Carrive

Audiovisual-based hypermedia authoring: using structured representations for efficient access to AV documents

In this article we introduce the notion of audiovisualbased hypermedia authoring systems, i.e. systems mainly using documents from digital audiovisual (AV) archives as a source for hypermedia authoring. After showing that traditional hypermedia models are ill-designed for specific constraints implied by such systems, we propose a change of approach. We present a model based on formal structured representations of the content of documents as it is done in the field of structured documents. Since a specific mode1 for the representation of AV content is needed. we introduce Audiovisual Event Description Inter

Representing temporal musical objects and reasoning in the MusES system

ace (AEDI), which provides a model for the description of AV documents. and an XML-based syntax for the exchange of such descriptions. We describe AEDIs main concepts, how it can be related to a formal specification of the domain knowledge -called onrofogy - which allows efficient dynamic hyperlinking among elements. Finally, we describe the implementation of this mode1 for the production of AV based hypermedia at INAs production department.

Modeling television schedules for television stream structuring

We describe a representation framework for temporal objects and reasoning in the MusES system. The framework is claimed to be use-neutral within the context of tonal music. It is based on the use of an object-oriented programming language, Smalltalk, and makes full use of two main representation mechanisms: class inheritance and delegation. We describe the kernel of the representation framework and explain the main design choices. We show how the kernel can be used and extended to represent important temporal concepts related to tonal music. The framework is validated by the realization of two substantial applications: an automatic analyser for chord sequences, and a simulator of jazz improvizations.

Conceptual Indexing of Televison Images Based on Face and Caption Sizes and Locations

TV stream structuring is very important for huge archives holders like the French National Institute, the BBC or the RAI, because it is the first necessary step to describe the various telecasts broadcast on various channels. It is also necessary for television ratings in order to isolate the telecasts that must be rated. One can think this structuring is a simple alignment of the TV guides on the stream itself. But TV guides present in average only 35% of the telecasts that represent in average less than 18 hours by day. We propose in this article a method to predict TV schedules by modeling the past ones in order to boil down the television stream structuring problem to a simple alignment problem.

A Language for Audiovisual Template Specification and Recognition

Indexing videos by their image content is an important issue for digital audiovisual archives. While much work has been devoted to classification and indexing methods based on perceptual qualities of images, such as color, shape and texture, there is also a need for classification and indexing of some structural properties of images. In this paper, we present some methods for image classification in video, based on the presence, size and location of faces and captions. We argue that such classifications are highly domain-dependent, and are best handled using flexible knowledge management systems (in our case, a description logics).

Using Description Logics for Indexing Audiovisual Documents

We address the issue of detecting automatically occurrences of high level patterns in audiovisual documents. These patterns correspond to recurring sequences of shots, which are considered as first class entities by documentalists, and used for annotation and retrieval. We introduce a language for specifying these patterns, based on an extension of Allens algebra with the regular expression operator +, which denotes an iteration of arbitrary length. We propose a formulation of this pattern language using the constraint satisfaction framework, in which templates are represented as constraint problems. We propose an efficient representation of domains (all subsequences of a given graph) and filtering methods for the Allen constraints. We illustrate the resulting system on a corpus of real world news broadcast examples.