一种改进粒子群分类算法的图像检索  

Image Retrieval Based on Improved Particle Swarm Optimization Classification Algorithm

作　　者：刘尚旺[1,2] 杨冉[1] 尚新闻[1] 崔艳萌 

机构地区：[1]河南师范大学计算机与信息工程学院,河南新乡453007 [2]河南师范大学,“智慧商务与物联网技术”河南省工程实验室,河南新乡453007

出　　处：《小型微型计算机系统》2016年第7期1624-1628,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(U1304607)资助;河南省高等学校重点科研项目(15A520080)资助;河南师范大学博士科研启动基金项目(qd12138)资助

摘　　要：传统基于内容的图像检索算法计算量大、精度低.为克服这些缺陷,在传统算法基础上,提出一种将图像匹配方法转化为根据模板图像数据,对目标进行锁定的方法:首先,将模板图像和源图像分区并取灰度直方图信息;然后,将寻找模板图像最相似的的问题转化成通过粒子群优化进行分类问题;最后,通过对相似度大的图像进行精确匹配得出最相似的图像.实验结果表明:基于粒子群和新分类算法的图像检索算法,能够在源图像数据库中快速匹配出相似的图像组,平均检索精度和查全率分别达到88.0%和93.3%,平均检索用时仅有3.48s.Traditional content-based image retrieval algorithms usually have large calculation and lowprecision. Aiming at solving these problems,the method of image matching was transformed to locate object according to template images. Firstly,extracting the histogram information of template image and images in database; Secondly,replacing the model of searching the similar images using template with the model of dynamic classification by PSO; Thirdly,the most similar class was selected by matching with the images having the large similarity. The experimental results showthat the proposed algorithm based on PSO and the novel classification could quickly find out a similar image groups from the original image database,and the average retrieval precision and recall rates reached88% and 93. 3%,respectively,and the average time consuming was only 3. 48 s.

关 键 词：分区 图像检索 粒子群优化算法 匹配 分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]