摘要
目前,基于内容的图像数据库检索己成为图像检索研究的主流,核心是图像相似性检索。其遇到的主要问题是检索的准确性需要进一步提高。由于免疫算法具有长期记忆和学习的能力,非常适合对用户的反馈信息进行长期的学习来提高系统对用户语义的理解能力。本文利用免疫算法的优点提出了基于免疫算法的图像检索模型并实现了一个原型化检索模型。对10000张corel图像库进行试验表明,相对于传统方法检索精度有了很大的提高。
At present CBIR is a core technique of image database system. The main obstacle facing content-based image retrieval is that the retrieval effectiveness is not satisfiable. Since immune algorithm has ability of lerning and memorying in longterm and in keeping with learning customer feedbacking information,it can improve the recognition of customers＇semantic target for system. In this paper using excellence of immune algorithm we have proposed a new relevant feedback model based on immune algorithm and implemented a archetypal model.Experimental results on 10 000 images show that this approach can improve the retrieval accuracy apparendy.
出处
《微计算机信息》
北大核心
2007年第04X期279-280,198,共3页
Control & Automation
基金
山西省留学人员科研资助项目(2004-26)
关键词
基于内容的图像检索
相关反馈
免疫算法
content-based image retrieval, relevance feedback, immune algorithm
