摘要
为了解决传统图像检索算法低效和耗时的缺点,提出一种基于PCA哈希的图像检索算法。通过结合PCA与流形学习将原始高维数据降维;然后通过最小方差旋转得到哈希函数和二值化阈值,进而将原始数据矩阵转换为哈希编码矩阵;最后通过计算样本间汉明距离得到样本相似性。在三个公开数据集上的实验结果表明,提出的哈希算法在多个评价指标下均优于现有算法。
In order to solve the inefficiency and time-consuming of traditional image retrieval algorithms,this paper proposed an image retrieval algorithm based on PCA hash.Specifically,by combining PCA and manifold learning,it reduced the dimensionality of the original high-dimensional data,and then obtained hash function and the binarization by minimum variance rotation.Then it converted the raw data matrix to a hash coded matrix.Finally,obtained the sample similarity by calculating the Hamming distance between samples.The experimental results on three public datasets show that the proposed hash algorithm outperforms the existing algorithms under multiple evaluation criteria.
作者
苏毅娟
余浩
雷聪
郑威
李永钢
Su Yijuan;Yu Hao;Lei Cong;Zheng Wei;Li Yonggang(College of Computer&Information Engineering,Guangxi Teachers Education University,Nanning 530023,China;Guangxi Key Labora-tory of Multi-source Information Mining&Security,Guangxi Normal University,Guilin Guangxi 541004,China)
出处
《计算机应用研究》
CSCD
北大核心
2018年第10期3147-3150,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(61672177
61573270)
国家"973"计划资助项目(2013CB329404)
广西自然科学基金资助项目(2015GXNSFCB139011
2015GXNSFAA139306)
广西研究生教育创新计划资助项目(XYCSZ2017064
XYCSZ2017067
YCSW2017065)
关键词
哈希
图像检索
主成分分析
流形学习
hashing
image retrieval
principal component analysis
manifold learning