摘要
针对传统石材检测方法实时性差、精度低与劳动强度大的缺点,提出了一种基于视觉的缺陷测方法,首先以石材轮廓的矩包络线为基准,利用近邻搜寻算法确定尺寸测量的角点,欧式距离衡量尺寸参数。然后采用分块策略生成对应的局部多特征映射矩阵,并将其元素与训练的标准参数进行对比,搜寻潜在缺陷块,结合邻域信息合并潜在块确定缺陷位置。最后应用多特征数据建立支持向量机(SVM,Support Vector Machine)的缺陷类别预测模型,实验表明该方法具有较好的检测效果,具有重要的应用价值。
Since the traditional method of stone detection still has shortcoming in real-time,low accuracy and high labor,a detection method for stone based on machine vision is proposed.Firstly,using rectangular envelope of stone as a benchmark,and adopting the nearest neighbor algorithm to acquire corner of measurement,calculating size parameter by Euclidean distance.Then Getting local feature mapping matrix through corresponding block strategy and comparing feature cell with trained parameters to search potential defects block,then combining neighborhood information to merge potential block to acquire the position of defect.Finally,establishing support vector machine(SVM)to predict defect classification by multi-features.The experimental results show that the method has good detection effect and high real-time performance.
作者
邓拥军
周向
DENG Yong-jun;ZHOU Xiang(Hubei University of Technology Engineering and Technology College,Hubei Wuhan 430068,China;Mechanical and Electrical Engineering Institute,Wuhan Technical College of Communications,Hubei WuHan 430000,China)
出处
《机械设计与制造》
北大核心
2019年第12期150-154,共5页
Machinery Design & Manufacture
基金
湖北省科技厅自然科学基金重点项目(2014CFA528)
武汉交通职业学院校级课题(Y2016011)
作者简介
邓拥军,(1971-),男,湖北武汉人,本科,工程师,主要研究方向:机械设计及自动化控制;周向,(1981-),女,湖南浏阳人,硕士研究生,讲师,主要研究方向:机器视觉及自动化检测技术。
