The purpose of the project of Object-based Surveillance Video Retrieval System is to extract foreground object by segmentation, and then describe extracted objects by feature descriptors. It could be used to analyze video semantically, for example, categorize objects into human or vehicle. It could also support the content-based video retrieval system to enhance surveillance purpose.
A digital video surveillance system that supports translation of conceptual query to semantic description for video searching. Real-time indexing and content-based distributed video retrieval techniques are developed in this project.
The objective of this project is to develop new and enabling semantic video analysis techniques for video retrieval applications. We believe these new techniques can help to remedy the video retrieval problems found in video surveillance systems and the video search engine.