关键词: 基于内容的图像检索, 灰关联度, 相似度量度

Abstract: A new algorithm of clustering based on gray association degree is presented to develop a method which can solve the problems of K-means and similarity measurement in the practical application.The proposed method introduces the principle of gray association and clustering based on gray association，and the corr-esponding algorithms is designed by VC++6.0.It is proved by experiment that the method not only overcomes the defects of the Euclidean distance on some extent，but also reflects the impact of inspected attribute.The method is applied to the rotary kiln flame image retrieval and the results demonstrate its usefulness and effectiveness，it can improve the speed of image retrieving.

Key words: Content-Based Image Retrieval（CBIR）, gray association rule, similarity measurement