陕西科技大学学报2018，Vol.36Issue(2)：147-153,7.
基于SVM的流程图像角点分类
Corner classification of flowchart based on SVM
摘要
Abstract
Existed flowchart recognition works failed to deal with touched line-texts and bro-ken lines.Corners in flowchart images are not affected by touched line-texts and broken lines and can be used to recognize logical structure of flowcharts to enable automatic comprehen-sion of flowchart images.An approach was proposed to automatically detect and classify the corners in flowchart images.Firstly,typical corner types are identified and a corner-based structural semantic model of flowchart is defined through analyzing flowchart structures from the perspective of corners.Secondly,the structural layer of a flowchart image is extrac-ted using the area of connected components and typical corner detectors are used synthetical-ly to identify corners in flowchart images.Finally,grid features and peripheral features of the neighborhood of corners are extracted for training a SVM-based corner classifier in a super-vised learning process.The corner classifier is validated using a public dataset from CLEP-IP.The overall corner recognition rate is 91.6%.关键词
流程图/角点检测/特征提取/SVM/角点分类Key words
flowchart/corner detection/feature extraction/SVM/corner classification分类
信息技术与安全科学引用本文复制引用
孙连山,张沙沙,侯涛,赵晓..基于SVM的流程图像角点分类[J].陕西科技大学学报,2018,36(2):147-153,7.基金项目
国家自然科学基金项目(61601271) （61601271）
陕西省教育厅专项科研计划项目(17JK0087) （17JK0087）
