摘要
针对传统单机视频检索效率低下的问题,设计了一种基于Hadoop的分布式视频车辆检索方法。该方法首先将视频切割成多个分块,然后利用MapReduce和FFMPEG在分布式环境下提取视频帧,再通过车牌识别算法检索其中出现的车牌号码,最后计算出车辆在视频中出现的时间。实验结果表明,与单机检索方式相比,该方式具有更强的数据处理能力和更高的检索效率。
In order to solve the problem that the Singe Operating System has the low video retrieval efficiency, in this paper, a method of distributed video vehicle retrieval based on Hadoop is proposed. Firstly, the videos are split up into video blocks. Secondly, using MapReduce and FFMPEG extract the video frames based on distributed environment. Subsequently, retrieve the license numbers appearing in the videos via the License Plate Recognition Algnrithm. Finally, the time when the vehicles appear in the video is realized. Comparing with the method of Singe-Retrieval, the test shows that this new method has more powerfnl capability for data processing and higher efficiency for retrieving.
出处
《电视技术》
北大核心
2015年第22期95-99,共5页
Video Engineering
基金
国家自然科学基金项目(61371107
61261017)
广西自然科学基金项目(2013GXNSFAA019334)
