【摘要】 提出了一种基于目标区域的图像综合特征检索方法。根据颜色、纹理特征的综合向量采用改进的K均值聚类算法把每幅图像分成若干个区域,并且利用综合特征表示和检索分割后的区域,该方法在区域匹配时考虑了不同特征在重要程度上的差别。为了改善检索性能,系统中引入了快速索引和相关反馈技术。更多还原

【Abstract】 This paper introduces a region-based image integrative feuture retrieval system.In this system,each image is segmented into several regions according to the integrative vector of color and texture feature by improved k-means clustering algorithm.The segmented regions are represented and retrieved using integrative features,and considering the difference in the importance of different features during matching the regions.In order to improve the retrieval performance,the fast indexing and relevance feedback technique are introduced.更多还原