现阶段基于内容的图像检索技术分析

Analysis to Development of Content-Based Image Retrieval

主要讨论了目前基于内容的图像检索技术的两大热点研究方向,即基于底层特征向量的提取算法(分别为颜色、纹理和形状特征)和基于相关反馈机制(分别为基于空间变换、基于统计概率、基于聚类分析和基于机器学习的模型).分析它们使用的典型方法及优缺点,并对未来的发展方向做了展望.

Two hot problems focusing on content-based image retrieval were discussed in this paper,and one of based features extracted of low-level for retrieval,including color,texture,and shape;another is based on relevance feedback,such as model of space,statistics,cluster and machine leaning. Their advantages and disadvantages are described. Finally,some promising research directions are suggested.