基于K-means的深度跨模态哈希量化优化方法  

K-means Based Quantitative-Optimization Method for Deep Cross-Modal Hashing

作　　者：吴家皋[1,2] 杨璐 翁玮薇 刘林峰 WU Jiagao;YANG Lu;WENG Weiwei;LIU Linfeng(School of Computer,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;Jiangsu Key Laboratory of Big Data Security&Intelligent Processing,Nanjing 210023,China)

机构地区：[1]南京邮电大学计算机学院,南京210023 [2]江苏省大数据安全与智能处理重点实验室,南京210023

出　　处：《南京航空航天大学学报》2021年第5期684-691,共8页Journal of Nanjing University of Aeronautics & Astronautics

基　　金：国家自然科学基金(41571389,61872191)资助项目。

摘　　要：互联网应用的普及使得多模态数据快速增长,跨模态检索技术已成为相关领域的关键技术之一。针对现有跨模态哈希算法存在的网络结构和量化方法等方面的问题,本文在新的深度跨模态哈希检索模型之上,提出了一种基于K-means的深度跨模态哈希量化优化方法(K-means-based quantitative-optimization for deep crossmodal hashing,KQDH)。该方法通过K-means聚类算法对多模态数据特征向量分类,并通过集体量化方式来控制量化误差,使得哈希码更好地表示出多模态特征。实验结果表明,该方法能在多模态数据之间保持相似性并最大程度地捕获语义信息,从而提高跨模态检索的准确性和效率。Multi-modal data are growing rapidly with the popularity of Internet applications,and cross-modal retrieval technology has become one of the key technologies in related research areas,where the cross-modal hash algorithm has been paid more and more attention because of its simplicity and efficiency recently.Due to the problems of existing algorithms in the network structure and quantization method,based on a new deep cross-modal hash retrieval model,a K-means based quantitative-optimization method for deep cross-modal hashing(KQDH)is proposed,which classifies the feature vectors of multi-modal data by K-means clustering algorithm,controls the quantization error by the collective quantization method,and makes the hash code better represent the multi-modal features.Experiments show that the proposed method can preserve the similarity between multi-modal data and capture semantic information to the greatest extent,and improve the accuracy and efficiency of cross-modal retrieval.

关 键 词：多模态 哈希算法 聚类算法 特征向量 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]