首页 | 本学科首页   官方微博 | 高级检索  
     

基于改进遗传算法的商标多特征权值分配
引用本文:王晓翠,刘建平. 基于改进遗传算法的商标多特征权值分配[J]. 计算机工程与设计, 2008, 29(8): 2038-2040
作者姓名:王晓翠  刘建平
作者单位:浙江理工大学,信息电子学院,浙江,杭州,310018
摘    要:提出一种改进的遗传算法用于解决商标图像检索中的多特征权值分配问题.针对以往的传统分配方法所存在的不足,提出了利用改进的遗传算法来分配权值.主要方法是将遗传算法中的轮盘赌选择与最优解保存法结合,并将染色体基因重新排列后进行交叉.把遗传算法用于图像检索系统的多特征权值分配,并将得到的特征权值用于商标图像检索系统,目的是为了提高检索系统的效率和精度,而且通过这种方法得到的权值检索出来的结果也比较符合人的视觉感官.

关 键 词:改进的遗传算法  早熟收敛  轮盘赌选择法  商标图像检索  多特征
文章编号:1000-7024(2008)08-2038-03
修稿时间:2007-05-23

Multiple-feature weight assignment for trademark based on revised genetic algorithm
WANG Xiao-cui,LIU Jian-ping. Multiple-feature weight assignment for trademark based on revised genetic algorithm[J]. Computer Engineering and Design, 2008, 29(8): 2038-2040
Authors:WANG Xiao-cui  LIU Jian-ping
Affiliation:WANG Xiao-cui,LIU Jian-ping(Department of Information , Electronics,Zhejiang Sci-Tech University,Hangzhou 310018,China)
Abstract:A revised genetic algorithm is proposed to solve the multiple-feature assignment in the trademark retrieval system.To overcome the deficiency in the traditional assignment,a revised genetic algorithm is proposed.The revised methods are the combination of Roulette wheel selection and the best solution conservation and the crossover of rearranged chromosomes.The new method is used in the trademark retrieval system and the efficiency and precision of the system is improved.The results also accord with the huma...
Keywords:revised genetic algorithm  premature convergence  roulette wheel selection  trademark retrieval  multiple-feature  
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号