摘要： 受现有灰色系统研究成果的启发，将灰色聚类方法应用于CBIR 的研究中，建立了CBIR 与灰色聚类的对应关系，提出了一种全新的基于灰色聚类的图像检索技术。该方法首先对用户主观性相似判断进行定性分析，确定若干个灰类；然后采用相应的训练算法建立各个灰类的白化函数；最后，采用灰色聚类方法对数据库图像进行聚类分析，得到检索结果集。这种方法既考虑了人类视觉感知的特点，同时又简化了问题的复杂度，从而使图像检索的效率与性能得以同时提高。

关键词: 灰色系统；灰色聚类；图像特征与表示；基于内容的图像检索

Abstract: Motivated by the research results of grey system theory, this paper proposes a novel grey clustering based approach for retrieving color images by applying the grey clustering method to image retrieval and mapping grey clustering decision to image retrieval. Firstly, to investigate the characteristics of human subjectivity of image similarity, the grey clusters are determined. Then, the albation functions are constructed by using the training algorithms. Finally, the grey clustering methods are applied to obtain the retrieval results. Since the proposed approach captures human perception subjectivity of image similarity, both the efficientness and performance are improved

Key words: Grey system; Grey clustering; Image feature and representation; Content-based image retrieval