The lifetime of this research project has seen two dramatic developments in the area of digital video coding. The first has been the progress of compression research leading to a factor of two improvement over existing standards, much wider deployment possibilities and the development of the new international ITU-T Recommendation H.263. The second has been a radical change in the approach to video content production with the introduction of the content-based coding concept and the addition of scene composition information to the encoded bit-stream. Content-based coding is central to the latest international standards efforts from the ISO/IEC MPEG working group. This thesis reports on extensions to existing compression techniques exploiting a priori knowledge about scene content. Existing, standardised, block-based compression coding techniques were extended with work on arithmetic entropy coding and intra-block prediction. These both form part of the H.263 and MPEG-4 specifications respectively. Object-based coding techniques were developed within a collaborative simulation model, known as SIMOC, then extended with ideas on grid motion vector modelling and vector accuracy confidence estimation. An improved confidence measure for encouraging motion smoothness is proposed. Object-based coding ideas, with those from other model and layer-based coding approaches, influenced the development of content-based coding within MPEG-4. This standard made considerable progress in this newly adopted content based video coding field defining normative techniques for arbitrary shape and texture coding. The means to generate this information, the analysis problem, for the content to be coded was intentionally not specified. Further research work in this area concentrated on video segmentation and analysis techniques to exploit the benefits of content based coding for generic frame based video. The work reported here introduces the use of a clustering algorithm on raw data features for providing initial segmentation of video data and subsequent tracking of those image regions through video sequences. Collaborative video analysis frameworks from COST 21 l qual and MPEG-4, combining results from many other segmentation schemes, are also introduced.