基于Hadoop的服饰图像存储与检索关键技术研究  

Research on key technology of clothing image storage and retrieval based on Hadoop

作　　者：郭飞[1] 詹炳宏[2,3] 刘刚[4] 

机构地区：[1]北京服装学院信息工程学院,北京100029 [2]北京服装学院艺术设计学院,北京100029 [3]北京市数字媒体与交互重点实验室,北京100029 [4]河南工业大学信息科学与工程学院,郑州450001

出　　处：《计算机应用研究》2014年第4期1086-1089,1126,共5页Application Research of Computers

基　　金：国家科技支撑计划资助项目(2013BAH41F00);国家自然科学基金资助项目(61174056);北京市数字与交互重点实验室资助项目(KF2012-15)

摘　　要：针对服饰图像都是小文件的特性,提出了一种HDFS和传统关系型数据库相结合的服饰图像及其特征数据的存储结构设计方法,实现了海量图像信息的快速存储和读取;改进了多特征点图像特征提取和匹配算法,并基于Map/Reduce框架实现了基于多特征的服饰图像数据分布式检索。实验结果表明,该方法能够均衡系统负载,提高资源利用率,扩展性强,有效地降低了海量服饰图像检索时间,是一种高效的服饰图像存储和检索的方法。According to the clothing image characteristics are small file, this paper proposed a storage structure design method of clothing image and their feature data, which combined HDFS and traditional relational database. It achieved rapidly storage and retrieval for massive image information. Then it improved multi-feature point image feature extraction and matching algo- rithms, and implemented the clothing image data distributed multi-feature retrieval based on the Map/Reduce platform. The experimental results demonstrate that this method can balance the system load, improve resource utilization and it has strong expansibility, effectively reduces the retrieval time of massive clothing image. It is a efficient storage and retrieval method for clothing image.

关 键 词：Hadoop云存储 小文件 基于内容的图像检索 服饰图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]