基于HBase的监控视频中目标图片的存储系统  

Storage System of Target Image in Surveillance Video Based on HBase

作　　者：李洁[1] 史云静[1] 郑海波[1] 韩小萱[1] 朱秀昌[1] 

机构地区：[1]南京邮电大学江苏省图像处理与图像通信重点实验室,江苏南京210003

出　　处：《计算机技术与发展》2015年第10期54-58,共5页Computer Technology and Development

基　　金：国家自然科学基金资助项目(61071091)

摘　　要：目标图片作为监控视频中重要的信息,在视频检索应用中起到重要的作用,同时随着监控视频数量的增长,目标图片的数量也是海量的,如何有效地存储这些目标图片成为研究热点。HBase是一种面向列的非关系型数据库,主要提供对海量数据的实时和随机的读写访问。文中通过分析监控视频中海量目标图片的特点和存储需求,以及对HBase的存储架构进行研究的基础上,提出一种基于HBase的海量监控视频中目标图片的高效存储方案。该方案完成了海量监控视频中目标图片的存储表和检索方案设计;同时,通过设置系统参数、预建Region、多线程并发写入实现了存储系统的写入优化。实验结果表明,该存储架构满足监控视频中海量目标图片的存储和检索要求,具有良好的水平扩展性,同时提高了写入效率。As the important information in surveillance video, target picture plays an important role in video retrieval applications, and with the growth in the number of surveillance videos, the number of target picture is massive, so how to effectively store these target images has became a hot topic. HBase is a column-oriented non-relational database, which provides the vast amounts of data in real time and random read and write access. By analyzing the target image features in surveillance videos and the massive storage architecture of HBase, propose an efficient storage solution of the target image in surveillance video based on HBase. This method has completed the de- sign of storage table of the target image and the retrieval programs. Besides by setting the parameters of HBase, pre-building regions, u- sing multi-threads to imported into HBase database. Finally,the experimental results show that this storage system has completed the re- quirements of storage and retrieval of massive images, which has a good level of scalability, and improves the efficiency of importing data into HBase.

关 键 词：HBASE 监控视频 目标图片 图片存储 

分 类 号：TP302.1[自动化与计算机技术—计算机系统结构]