以Hadoop平台为依托设计海量视频检索系统

Designing a massive video retrieval system based on the Hadoop platform

Hadoop系统具有四高优点,即高包容性、高可用性、高运行效率和高可靠性,目前应用领域极为广泛。以Hadoop平台为依托设计海量视频检索系统需满足快速高效检索的现实需求和强大的存储及分布式转码的功能需求。系统架构时需要以Hadoop平台的三大核心组件为依托,整个系统结构设计为六层,同时,依托H_u不变矩等先进算法提取目标图像特征进行相似度匹配,最终完成检索。

The Hadoop system has four advantages:high inclusivenes,high availability,high operating efficiency and high relia-bility.It is currently widely used.Designing a massive video retrieval system based on the Hadoop platform needs to meet the real needs of fast and efficient retrieval and the powerful storage and distributed transcoding functions.The system architecture needs to rely on the three core components of tlie Hadoop platform.The whole system structure is designed as six layers.At the same time,advanced image algorithms such as N invariant moment are used to extract the target image features for similarity matching,and f-nally the retrieval is completed.