|Title||Evaluation of Visual Features for Automatic Summarization of Video Sequences|
|Tutor||Dipl.-Ing. Martin Haller|
|Professor||Dr.-Ing. Thomas Sikora|
|Abstract||With the increase of image and video data, it becomes a challenge to manage this huge amount of visual data. Here visual similarity analysis is a key technique for exploring and determine visual similarity structure of unstructured visual data, especially for video summarization and content-based image retrieval (CBIR).|
This thesis presents an exhaustive evaluation of visual features used for visual similarity analysis. The evaluation considers the feature space as well as clustering scenario for video summarization using TRECVid 2007 BBC rushes test videos.
|Key words||MPEG-7, PICSOM, cluster analysis, feature evaluation, BIF SFS methods|