欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2013, Vol. 30 ›› Issue (1): 106-111.DOI: 10.7523/j.issn.1002-1175.2013.01.016

• 计算机科学 • 上一篇    下一篇

一种基于分布式LSH的海量视频快速检索方法

曹海傧, 朱明, 冯伟国   

  1. 中国科学技术大学自动化系网络传播系统与控制安徽省重点实验室, 合肥 230027
  • 收稿日期:2012-01-04 修回日期:2012-03-21 发布日期:2013-01-15
  • 通讯作者: 朱明
  • 基金资助:

    国家科技支撑计划课题(2011BAH11B01)和国家自然科学基金(60975045)资助

Fast search method for large-scale video based on distributed LSH

CAO Hai-Bin, ZHU Ming, FENG Wei-Guo   

  1. Key Lab of Network CommunicationSystem & Control of Anhui Province, Department of Automation, University of Science and Technology of China, Hefei 230027, China
  • Received:2012-01-04 Revised:2012-03-21 Published:2013-01-15

摘要:

针对大规模视频数据库,提出一种快速的视频相似性检索方法.该方法采用基于内存约束的分布式哈希结构,称为MD-LSH,有效地解决了快速与可扩展性问题.首先从视频序列中提取出特征向量集合,然后使用MD-LSH对高维特征向量建立索引,根据返回的相似帧集合对相关的视频进行相似度计算,最后排序得到相似视频的查询结果.实验表明,该方法对大规模视频的快速检索是非常有效的.

关键词: 基于内容的召回, 分布式LSH, 海量视频, 快速检索

Abstract:

we propose a fast similarity-based video retrieval method for large-scale video database. This method solves scalability and efficiency problems. First, we extract feature vector set for every video and build index using a distributed memory hash structure, called MD-LSH. Secondly, we calculate the similarity of the relevant videos according to the returned similar frame sets. Finally, we return the sorted similar video list as the query results. Experiments show the effectiveness of the proposed method for large-scale video fast retrieval.

Key words: content-based retrieval, distributed LSH, large-scale video, fast search

中图分类号: