摘要
基于遥感图像的光谱特征和纹理特征,提出图像关联规则的定义,通过对图像降阶预处理构造图像模板,采用关联规则挖掘算法对图像纹理的频繁模式进行挖掘,通过联合关联规则来表达纹理。试验表明,挖掘出的关联规则不仅能够表达规则纹理,而且能够较好表达随机纹理,据此可以进行纹理图像分割。
Based on the gray and texture feature of remote sensing image, this paper proposes the concept of association rule in image. By degrading the image, the mask of image can be constructed, and the frequent patterns of texture can be mined by the algorithm of association rules mining. The association rules can be combined to represent the texture. The experimental results validate that the combined association rules can represent the regular texture and represent the random texture perfectly. Therefore, the image segmentation can be accomplished by the combined association rules.
出处
《重庆邮电大学学报（自然科学版）》
2007年第4期438-441,449,共5页
Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)
基金
重庆市教育委员会科学技术研究项目资助(KJ060511)
关键词
数据挖掘
纹理
图像
关联规则
data mining
texture
image
association rule
