摘要
针对大数据数据库中图像索引中维度灾难问题,该文提出一种基于云的大规模图像检索技术,该方法创新性地将主成分分析法和二叉树引入到图像检索技术中,首先采用尺度不变特征变换和加速鲁棒特征描述符作为帧特征,面对大规模维度特征,将主成分分析法对帧特征进行降维,并使用二叉树表示降维后的特征,以加速研究阶段并减少存储空间,最终实现图像检索.实验表明:该文方法在降维70%的条件下,搜索精确率/召回率(Precision/Recall,PR)值能够达到传统方法20%降维条件下的PR值,并且在搜索时间上,该文方法与正常搜索相比,搜索速度得到30%~50%的提升.
In order to solve the problem of dimension disaster in the image index in large data database,a large scale image retrieval technology based on cloud has been proposed in this paper.In the method,principal component analysis and binary trees have innovatively been introduced into image retrieval technology.First,scale invariant feature transform and speeded up robust features descriptor are used as the frame features.In the face of large-scale dimension features,the principal component analysis method is used to reduce the dimension of the frame feature,and a binary tree is used to represent the features after the dimension reduction to accelerate the research phase and reduce the storage space.Finally,image retrieval is realized.Experiments show that under the condition of reducing the dimension by 70%,the PR value of this method can reach the PR value under the traditional method of 20% dimensionality reduction.Compared with normal search,the search speed of this method is increased by 30%~50%.
作者
周雪梅
潘多
ZHOU Xue-mei;PAN Duo(Department of Information Engineering,Sichuan Technology and Business College,Dujiangyan Sichuan 611830,China)
出处
《西南师范大学学报(自然科学版)》
CAS
北大核心
2019年第7期57-62,共6页
Journal of Southwest China Normal University(Natural Science Edition)
基金
教育部科技发展中心产学研创新基金课题(2018A03007)
关键词
大数据
大规模图像索引
主成分分析
二叉树
尺度不变特征变换
big data
large scale image indexing
principal component analysis
binary trees
scale invariant feature transform