基于内容的多媒体数据库管理系统的研究

Research of Multimedia Database Management System Based on Content Retrieval

通过分析数据库的数据模型,研究基于内容的多媒体数据库管理系统的构建方法及其功能框架,采用PL/SQL方式访问Oracle 8i数据库。为了提高管理系统的图像检索速度,提出了一种基于内容的图像检索算法,从聚类中心初值选取和分类中心的更新方面改进C-均值聚类算法,较好地解决了图像的分类问题。实验表明:使用该聚类检索算法,能对分类中心进行快速更新,有效地对图像进行聚类以及缩短检索时间,检索性能优于现有的C-均值聚类算法。

Through analyzing the data model of database, studies the construction method and its functional framework of the content- based multimedia database management system by means of accessing the Oracle 8i database through PL/SQL. To increase the speed of image retrieval of the management system, proposes a content - based irnage retrieval algorithm which aims at improving the C - means clustering algorithm by selecting initial value of the cluster center and updating the classified center. Results of the experiment show that the application of this clustering retrieval algorithm can update the classified center quickly, cluster the relevant images effectively, shorten the retrieval time, and its retrieval performance is superior to the existing C- means clustering algorithm.