面向云环境的图像高维特征索引框架

• 论文 •

面向云环境的图像高维特征索引框架

陈凤娟，丁贵广，朱妤晴

1.清华大学计算机科学与技术系，北京100084；2.清华大学软件学院，北京100084

出版日期:2011-08-15 发布日期:2011-08-25

Cloud-oriented image high-dimensional feature index framework

CHEN Feng-juan, DING Gui-guang, ZHU Yu-qing

1.Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China;2.School of Software, Tsinghua University, Beijing 100084, China

Online:2011-08-15 Published:2011-08-25

摘要/Abstract

摘要： 针对海量图像数据的高维特征索引和查询方法，设计了一个面向云环境的两阶段图像高维特征索引框架，并基于MapReduce机制进行了系统实现。提出了一种基于位置敏感哈希函数的两阶段索引框架，可有效支持高维特征索引的分布式创建；利用MapReduce计算机制，设计和实现了分布式索引构建和查询算法，并集成到非结构化数据管理系统中。实验结果表明，该索引框架的查询速度随着数据规模不断增大呈亚线性增长。

关键词: 高维特征索引, 分布式索引, 位置敏感哈希算法, 基于内容的图像检索

Abstract: Aiming at high-dimensional feature indexing and searching method of mass images data in the cloud, a two-phase cloud-oriented image high-dimensional feature indexing framework was designed and was implemented based on MapReduce mechanism. A based Image high-dimensional Feature indexing Framework (LIFF) based on Locality Sensitive Hash (LSH) function was proposed to effectively support the distributed creation of massive high-dimensional feature data in the cloud. Distributed indexing and searching algorithm based on MapReduce framework was designed and implemented, which was integrated into laSQL Unstructured Data Management System (LaUD-MS). Experimental results showed that query speed on the LIFF index was sub-linear growth with the data scale increased constantly.

Key words: high-dimensional indexing, distributed indexing, locality sensitive Hash algorithm, content-based image retrieval, cloud computing

中图分类号:

TP319.4

陈凤娟，丁贵广，朱妤晴. 面向云环境的图像高维特征索引框架[J]. .

CHEN Feng-juan, DING Gui-guang, ZHU Yu-qing. Cloud-oriented image high-dimensional feature index framework[J]. .