摘要
为实现化纤长丝断头自动化巡检,提高企业生产效率,提出一套基于机器视觉的化纤长丝断头检测系统。检测系统通过工业相机采集长丝图像,并采用小波去噪和阈值分割算法进行图像增强处理;利用霍夫变换实现长丝主体与断头部分的形状分割;基于提取的多个形状特征,建立和训练径向基函数神经网络模型,实现对化纤长丝断头的检测和分类。选取150 tex/36 f规格的POY(pre-oriented yarn)长丝进行试验,结果显示该算法断头识别率超过95%,断头识别时间由人工检测的10^(2) s数量级提升到10^(-1) s数量级。
In order to realize the automatic inspection of chemical fiber filament breakage and improve the production efficiency of enterprises, a detection system of chemical fiber filament breakage based on machine vision was proposed. When the filament image was collected by industrial camera, the detection system used wavelet de-noising and threshold segmentation algorithm for image enhancement;Hough transform was used to realize the shape segmentation of filament body and broken end;based on the extracted shape features, the radial basis function neural network model was established and trained to realize the detection and classification of chemical filament broken end. The 150 tex/36 f pre-oriented yarn(POY) filament is selected for test. The results show that the recognition rate of broken ends of the algorithm exceeds 95%, and the recognition time of broken ends increases from the order of 10^(2)s for manual detection to the order of 10^(-1)s.
作者
陈思俊
陈振中
CHEN Sijun;CHEN Zhenzhong(College of Mechanical Engineering,Donghua University,Shanghai 201620,China)
出处
《东华大学学报（自然科学版）》
CAS
北大核心
2022年第1期58-65,共8页
Journal of Donghua University(Natural Science)
基金
上海市自然科学基金(19ZR1401600)
中央高校基本科研业务费专项资金(18D110316)
国家自然科学基金(51905492)
关键词
长丝断头检测
机器视觉
图像处理
径向基函数神经网络
filament broken end detection
machine vision
image processing
radial basis function neural network
作者简介
通信作者:陈振中,男,博士,副研究员,研究方向为机器视觉、模式识别和机器学习等,E-mail:zhenzh.chen@dhu.edu.cn。
