CAJ | 학술논문

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。
목표도편작위감공시빈중중요적신식，재시빈검색응용중기도중요적작용，동시수착감공시빈수량적증장，목표도편적수량야시해량적，여하유효지존저저사목표도편성위연구열점。 HBase시일충면향렬적비관계형수거고，주요제공대해량수거적실시화수궤적독사방문。문중통과분석감공시빈중해량목표도편적특점화존저수구，이급대HBase적존저가구진행연구적기출상，제출일충기우HBase적해량감공시빈중목표도편적고효존저방안。해방안완성료해량감공시빈중목표도편적존저표화검색방안설계；동시，통과설치계통삼수、예건Region、다선정병발사입실현료존저계통적사입우화。실험결과표명，해존저가구만족감공시빈중해량목표도편적존저화검색요구，구유량호적수평확전성，동시제고료사입효솔。
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.