Image Retrieval Approach Based on Intuitive Fuzzy Set Combined with Genetic Algorithm  

作　　者：王潇茵 徐卫华 胡昌振 

机构地区：[1]Lab of Computer Network Defense Technology ,Beijing Institute of Technology [2]Automation Station of Army Aviation Department of the General Staff

出　　处：《Journal of Beijing Institute of Technology》2009年第1期60-64,共5页北京理工大学学报（英文版）

基　　金：Sponsored by the Ministerial Level Foundation(20061823)

摘　　要：Aiming at shortcomings of traditional image retrieval systems, a new image retrieval approach based on color features of image combining intuitive fuzzy theory with genetic algorithm is proposed. Each image is segmented into a constant number of sub-images in vertical direction. Color features are extracted from every sub-image to get chromosome coding. It is considered that fuzzy membership and intuitive fuzzy hesitancy degree of every pixel＇s color in image are associated to all the color histogram bins. Certain feature, fuzzy feature and intuitive fuzzy feature of colors in an image, are used together to describe the content of image. Efficient combinations of sub-image are selected according to operation of selecting, crossing and variation. Retrieval results are obtained from image matching based on these color feature combinations of sub-images. Tests show that this approach can improve the accuracy of image retrieval in the case of not decreasing the speed of image retrieval. Its mean precision is above 80 %.Aiming at shortcomings of traditional image retrieval systems, a new image retrieval approach based on color features of image combining intuitive fuzzy theory with genetic algorithm is proposed. Each image is segmented into a constant number of sub-images in vertical direction. Color features are extracted from every sub-image to get chromosome coding. It is considered that fuzzy membership and intuitive fuzzy hesitancy degree of every pixel＇s color in image are associated to all the color histogram bins. Certain feature, fuzzy feature and intuitive fuzzy feature of colors in an image, are used together to describe the content of image. Efficient combinations of sub-image are selected according to operation of selecting, crossing and variation. Retrieval results are obtained from image matching based on these color feature combinations of sub-images. Tests show that this approach can improve the accuracy of image retrieval in the case of not decreasing the speed of image retrieval. Its mean precision is above 80 %.

关 键 词：intuitive fuzzy genetic algorithm color histogram image retrieval 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] O159[自动化与计算机技术—计算机科学与技术]