Output details

Shot Boundary Detection in MPEG Videos Using Local and Global Indicators

<23>This research was ranked first of 15 teams (including AT&T and Yale) participating in TRECKVID, a worldwide annual video processing research contest organized by the American National Institute of Standards and Technology, on the task of shot boundary detection, a fundamental technology for computerized video content analysis and management. Compared with existing approaches, this work contributes significantly to shot boundary detection through (i) direct detection of boundaries in compressed domain, conferring low computing complexity and high processing speed; (ii) the successful introduction of likelihood and multi-level content indicators (both local and global) to measure content similarity effectively within the neighbourhood.