摘要
图像挖掘技术与关联规则的结合在网络数据索引中占据了先机,但一些功能弊端不可避免。在这样的背景下,对关联规则的编码、特征点排列和运算方法进行改进。改进关联规则将网络数据集合转化为布尔矩阵,实行列内积运算,保留矩阵内大于或等于图像特征最小支持度的逻辑,挖掘出高频特征集合。设计基于改进关联规则的图像挖掘系统,系统包含数据采集、预处理、数据库和图像挖掘四个结构层,给出具备去噪、分压和滤波功能的图像预处理电路,并介绍了图像信息数据库结构,最后通过实验证明系统可进行高效率的图像挖掘,并且图像区分度大。
Abatract： The combination of image mining technology and association rules plays an important role in network data index, but still can＇t avoid some functional disadvantages. In this context, the coding, feature point arrangement and operating method of association rules are improved. The network data set is converted into Boolean matrix by means of the improved association rules to carry out the column inner product operation and reserve the logic value which is greater than or equal to that of image feature minimum support, so as to mine the high-frequency feature set. The image mining system based on improved association rules was designed. The system includes the structural layers of data acquisition, preprocessing, database and image mining. The image preprocessing circuit with the functions of denoising, voltage distribution and filtering is given. The structure of the image information database is introduced. The experimental results prove that the system can mine the image effectively, and has high image discrimination.
出处
《现代电子技术》
北大核心
2017年第16期109-111,116,共4页
Modern Electronics Technique
基金
吉林省自然科学基金研究项目资助(163501320442)
吉林省教育厅课题项目(吉教科合字[2015]第353号)
关键词
关联规则
图像挖掘
布尔矩阵
内积运算
图像信息数据库
association rule
image mining
Boolean matrix
inner product operation
image information database
