摘要
铜、铝作为报废汽车拆解破碎得到的主要有色金属,具有极高的市场价值,但因种类复杂和经过破碎导致的特性对分选造成困难。该文将机器视觉技术引入有色金属破碎料的快速识别中,通过机器视觉硬件获取不同种类的铜铝碎料局部纹理图360张,300张作为训练样本,60张作为测试样本,对所有样本提取样本和纹理共13项特征,利用随机森林对高维数据进行鉴别,平均识别准确率在95%以上,并选出了最优分类器和最优分类特征。该文研究的基于机器视觉的破碎料识别方法可以快速地进行特征提取和有效的数据降维,为批量进行破碎料分选提供了技术支持。
Copper and aluminum are the main non-ferrous metals which comse from dismantling and crushing of scrap cars,and have high market value,but the complexity of the types and the characteristics of the crushing make these difficult to select.This paper introduces the machine vision technology in the quick identification of non-ferrous crushing metal.Through machine vision hardware took a total of 360 figures of local texture,300 as the training samples,60 as test samples.All samples were extracted with 13 features of color and texture,then the random forest algorithm was used to identify these high-dime nsional datas.The average recognition accuracy is above 95%,the optimal classifier and most classification feature have also been picked out.In this paper,the method of recognition based on machine vision can quickly accomplish feature extraction and effective dimensionality reduction,,which provides technical support for batch sorting of non-ferrous crushing metal.
作者
许子鸣
胡志力
XU Zi-ming;HU Zhi-li(Hubei Key Laboratory of Advanced Technology of Automotive Components,Hubei Collaborative Innovation Center for Automotive Compoments Technology,Wuhan University of Technology,Wuhan 430070,China)
出处
《自动化与仪表》
2018年第3期92-96,共5页
Automation & Instrumentation
基金
湖北省重大科技创新计划项目(2015AAA014)
关键词
机器视觉
颜色特征
纹理特征
破碎料
随机森林
machine vision
color features
texture features
crushing metal
random forest(RF)
