图书馆图片检索系统的几点关键技术研究

基于内容的图像检索方法是目前图书馆查询大量图片资料的关键技术。本文对低层视觉特征提取、高维索引、相似性度量准则与相关反馈技术进行较深入的研究，提出了一种颜色特征提取和特征向量的索引方法，讨论了系统架构以及系统的编程实现，数据模型的操作和系统的性能测试。实验证明，本方法在一定程度上提高了检索的准确率和效率。

Study on Key Techniques of System of Image Retrieval of Library

The method of content-based image retrieval is now a key technology to effectively query a mass of image materials. In this paper, a lot research work has been done around four points: how to abstract low-level feature of vision, how to index in high-dimensional, how to adjust similarity measure norm and use relevant feedback technique. An index method of abstract the color-feature and feature vectors is proposed, discuss the construction and editor program of system, and the operating of data model and testing of system. Experiment shows that the method could increase index accuracy and efficiency.