文章基本信息

标题：Multi Features Content-Based Image Retrieval Using Clustering And Decision Tree Algorithm
本地全文：下载
作者：Kusrini Kusrini ; M. Dedi Iskandar ; Ferry Wahyu Wibowo 等
期刊名称：TELKOMNIKA (Telecommunication Computing Electronics and Control)
印刷版ISSN：2302-9293
出版年度：2016
卷号：14
期号：4
页码：1480-1492
DOI：10.12928/telkomnika.v14i4.4646
语种：English
出版社：Universitas Ahmad Dahlan
摘要：The classification can be performed by using the decision tree approach. Previous researches on the classification using the decision tree have mostly been intended to classify text data. This paper was intended to introduce a classification application to the content-based image retrieval (CBIR) with multi-attributes by using a decision tree. The attributes used were the visual features of the image, i.e. : color moments (order 1, 2 and 3), image entropy, energy and homogeneity. K-means cluster algorithm was used to categorize each attribute. The result of categorized data was then built into a decision tree by using C4.5. To show the concept in application, this research built an application with main features, i.e.: cases data input, cases list, training process and testing process to do classification. The resulting tests of 150 rontgen data showed the training data classification’s truth value of 75.33% and testing data classification of 55.7%.
其他摘要：The classification can be performed by using the decision tree approach. Previous researches on the classification using the decision tree have mostly been intended to classify text data. This paper was intended to introduce a classification application to the content-based image retrieval (CBIR) with multi-attributes by using a decision tree. The attributes used were the visual features of the image, i.e. : color moments (order 1, 2 and 3), image entropy, energy and homogeneity. K-means cluster algorithm was used to categorize each attribute. The result of categorized data was then built into a decision tree by using C4.5. To show the concept in application, this research built an application with main features, i.e.: cases data input, cases list, training process and testing process to do classification. The resulting tests of 150 rontgen data showed the training data classification’s truth value of 75.33% and testing data classification of 55.7%.
关键词：Image;Classification;Decision Tree;Clustering