基于兴趣点颜色形状特征的图像检索方法  

Image Retrieval Based on Color and Shape Features of Interest Points

作　　者：何姗[1] 郭宝龙[1] 洪俊标[1] 

机构地区：[1]西安电子科技大学机电工程学院,陕西西安710071

出　　处：《微电子学与计算机》2006年第3期40-42,46,共4页Microelectronics & Computer

摘　　要：提出了一种基于兴趣点的图像检索新方法ACSM(AnnularColorandShapeMoment),利用兴趣点的位置信息,提取兴趣点周围局部区域的颜色矩特征和区域形状不变矩特征。ACSM算法不仅对平移、旋转具有良好的鲁棒性,而且克服了传统颜色矩和形状不变矩没有位置信息的缺陷。实验结果表明,由于将颜色和形状特征有机结合,ACSM算法有效提高了图像检索的准确性。This paper presents a novel algorithm for image retrieval based on interest points, which utilizes the location information of the interest points, extracts the local color moment feature and shape invariant moment feature. This algorithm is not only robust to translation and rotation, but also avoids the drawbacks of losing the location information in color moment and shape invariant moment. Experimental results show that this algorithm effectively improves the image retrieval accuracy due to combining color and shape features.

关 键 词：图像检索 兴趣点 颜色矩 形状不变矩 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]