Lines of research

Research on video sequence analysis is mainly focused in the following lines:

  • Computer vision applied to video-surveillance: people detection, tracking and counting, abandoned/stolen objects, detection of camera maladjustements (out of focus, displacements, oclussions), etc.
  • Distributed video sequence analysis, multi-camera systems (calibration, fusion), etc.
  • Application of ontologies in the context of segmentation and object recognition.
  • Analysis of video sequences in compressed domain (MPEG-2, MPEG-4, H-264): object segmentation in mobile camera and dynamic background environments, temporal segmentation in shots and scenes, event detection, etc.
  • Identification and extraction of mid-level features from low-level analysis for annotation and semantic adaptation of video sequences purposes.

Within multimedia content adaptation, main lines of research include:

  • On-Line adaptation engines: generic systems modelling, open and scable approaches, content streaming, etc.
  • Schemes for content adaptation management, decision processes and terminal, user and network profiles (MPEG-7, MPEG-21).
  • Video sequences summarization: storyboard and video-skim creation following different approaches based on previous content analysis. Hierarchical approaches based on frames or GoPs over raw video, MPEG-2 and H-264. Real-time generation of video summaries on download time, personalized summaries generation, etc.
  • Content adaptation based on modality change (e.g. image to video adaptation).