改进的基于广义图像共生矩阵的图像检索方法  

An improved method for image retrieval based on generalized co-occurrence matrix

作　　者：洪清启[1] 王备战[1] 董槐林[1] 张磊[1] 陈兵[1] 

机构地区：[1]厦门大学软件学院,福建厦门361005

出　　处：《微型机与应用》2007年第S1期103-105,共3页Microcomputer & Its Applications

基　　金：985二期信息创新平台项目资助(0000-X07204)

摘　　要：综合考虑了传统灰度共生矩阵法与基于广义图像灰度共生矩阵法各自的优点,提出了改进的基于广义图像灰度共生矩阵的图像检索方法。新方法构造了广义图像四个方向的灰度共生矩阵,并提取四个共生矩阵的纹理参数进行检索。实验结果表明,新方法对图像的旋转及尺寸变化具有更好的检索性能。This paper integrates the advantages of the traditional co-occurrence matrix and generalized image co-occurrence matrix and proposes an improved method based on generalized image co-occurrence matrix. The new approach constructs four co-occurrence matrixes based on generalized image in four directions, and then it picks up the texture parameters of the four co-occurrence matrixes for image retrieval. The experimental results show that the new method has an advanced general performance and the retrieval effi...

关 键 词：基于内容的图像检索 灰度共生矩阵 广义图像 平滑图像 纹理特征 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]