Successive approximations for scalable content and content descriptors in multimedia applications

Interactive media aims to provide seamless access to digital content in a proactive and user centred manner. Regardless of the nature of the delivery channel, homogeneous or heterogeneous, and regardless of the user platform, media should be accessible efficiently and at optimum quality over the full available bandwidths and independently of terminal capabilities. In this paper, we argue that the development of efficient applications underpinning interactive media systems need to be based on successive approximations of content and metadata. The presented discussion relates to the content engineering side of media applications and describes how it can be used to easy, if not to achieve, the overall goal of a new generation of media applications. Three examples combining available technologies and ongoing developments to produce successive approximations of content and metadata are presented. Their application in the development of interactive multimedia systems is discussed and results of computer experiments aiming to demonstrate the suitability of these techniques are reported.