This seminar presents our recent contribution to RUSHES Project which is supported by EU FP6. The target of RUSHES Project is to develop a prototype for seamless indexing, searching and retrieval of content, especially applied to archives of raw media material, in order to ease in-house post-production in both professional and home environments.
Our contribution includes proposing and implementing new algorithms for video annotation, multi-modality analysis, and video summarisation. As a result, some technical details will be summarized as follows: (1) As an example of video annotation a new scheme namely iterative Random Sample Consensus (ARANSAC) scheme will be introduced for extracting planar surfaces from 2D image sequences. (2) To aid multimodality analysis in video retrieval, where audio and visual components are taken into account, a new feature detection algorithm is proposed for discriminating between speech and music based on the averaged cepstrum of the audio signals. (3) A content based dynamic video summarisation scheme is presented, which provides a short synopsis of a long movie by analysing colour and motion information of the original video.
Huiyu Zhou obtained his BEng degree in Radio Technology from the Huazhong University of Science and Technology of China, and an MSc in Biomedical Engineering from the University of Dundee of Scotland, respectively. He then received his PhD degree in Computer Vision from the Heriot-Watt University, Edinburgh, Scotland. He has worked in the Guangxi Medical University of China, Elscint Ltd. of Israel, University of Essex and University of London of United Kingdom. His research interests include computer vision, medical imaging, robotics and intelligent sensing systems. He has published widely in these areas. Currently, he is a research fellow at the Centre for Media Communications Research of Brunel University, United Kingdom, supported by European Commission under Grant FP6-045189-STREP (RUSHES).