介绍了一种基于H.264格式压缩的监控视频检索系统,该系统基于对象和语义,主要包括视频分析和视频检索两大模块。通过视频分析模块,对H.264压缩域视频解码并进行视频对象提取、分割、特征提取和对象匹配等分析过程,将得到的视频对象及特征信息存入至对象特征数据库;在视频检索模块端,用户可以通过输入语义信息或示例图片直接从特征数据库中进行视频检索查询,避免了视频的重复处理。实验表明,通过提出的系统可以快速有效地进行视频检索。
In this paper, a surveillance video retrieval system based on H.264 is introduced. The proposed system is object-semantic-based, and is mainly composed of two modules: video analysis and video retrieval. In the video analysis, the system decodes the H.264 files firstly, then extracts and segments video object from the decoded frames. Feature information which is extracted from video object is used to video matching. The obtained video objects and feature information are stored into feature database. During retrieval, the queried semanteme and/or snapshot input by user is eonverted to semantic information without re-processing the surveillance videos. The experiment shows that the user can retrieve the required videos effectively.