综合纹理特征和语义信息的医学图像检索  

Medical image retrieval based on texture features and semantic information

作　　者：王李冬[1] 邰晓英[1] 巴特尔[2] 赵杰煜[1] 

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211 [2]内蒙古自治区医院,内蒙古呼和浩特010017

出　　处：《计算机应用》2005年第10期2383-2386,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60472099)

摘　　要：提出一种结合图像分块纹理特征和语义信息的医学胸片图像检索方法。同时,介绍了颜色特征提取方法中的颜色相关图算法。据此,实现了一个图像检索原型系统,依据所设计的评价实验,将不同实验的检索结果进行了比较和分析。实验证明,结合图像分块纹理特征和语义信息的检索方法具有较好的检索效果。This paper presented a method of medical images retrieval about sternums based on texture features combining with semantic information. The texture features were calculated with gray level co-occurrence matrix in different partitions. This paper also introduced an algorithm about color correlogram. A prototype system was implemented. Meanwhile, three experiments were designed to evaluate the effectiveness of the approach. The comparison of the retrieval results show that the approach proposed in the paper is effective.

关 键 词：医学图像检索 灰度层共现矩阵 颜色相关图 语义信息 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]