摘要
针对传统BOW(Bag of Words)模型用于场景图像分类时的不足,通过引入关联规则的MFI(Maximum Frequent Itemsets)和Topology模型对其进行改进。为了突出同类图像的视觉单词,提取同类图像的MFI后,对其中频繁出现的视觉单词进行加权处理,增强同类图像的共有特征。同时,为了提高视觉词典的生成效率,利用Topology模型对原始模型进行分工并行处理。通过COREL和Caltech-256图像库的实验,证明改进后的模型提高了对场景图像的分类性能,并验证了其Topology模型的有效性和可行性。
Aimed at the defects of the traditional bag of words representation algorithm on the scene image categorization, an improved model, based on the maximum frequent itemsets （MFI） from association rule and the Topology model, is proposed. To highlight the visual words and fully express the common image characteristics in one category, the improved algorithm weights the frequent visual words after extracting MFI from one category. Meanwhile, by using the Topology model to make the progress of producing visual word dictionary more efficient, the original model can be distributed into each component by parallel operation. Experimental results based on COREL and Caltech-256 database demonstrate a better classification performance for scene image and shows the effectiveness and the feasibility of the Topology model.
出处
《南京邮电大学学报（自然科学版）》
北大核心
2016年第6期24-29,38,共7页
Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition
关键词
图像分类
BOW模型
MFI
TOPOLOGY
image categorization
bag of words （BOW） model
maximum frequent itemsets （ MFI ）
Topology
