摘要
为客观评价黑板纱线毛羽水平,将绕有纱线的黑板经扫描仪采集图像,通过二值化、形态学运算、局部阈值等处理,得到黑板毛羽图像和毛羽量像素点,提出基于图像处理技术的环锭纺纱线黑板毛羽 M 指数,探讨 M 指数与毛羽 H 值、毛羽根数之间的关系。将24种环锭纺纱线的 S 1+2 值和 M 指数分别输入到BP(back propagation)神经网络和RBF(radical basis function)神经网络中训练并预测毛羽 H 值。将预测毛羽 H 值与实际毛羽 H 值进行比较,结果表明,在预测精度上,BP网络模型的预测效果最好,RBF网络次之,多元线性回归模型预测效果最差。
In order to evaluate the quality of the blackboard yarn hairiness objectively, the image of blackboard yarn was acquired by a scanner. The result of blackboard yarn hairiness and the pixels of yarn hairiness were obtained through binarization, morphological calculation and local threshold. The Index M of ring spun yarn on the blackboard was proposed by the image processing. The relationship among index M , hairiness H value and the number of hairiness was discussed. The S 1+2 value and M index of the 24 kinds of ring spun yarns were input into the BP (back propagation) neural network and the RBF (radical basis function) neural network respectively to train and predict the hairiness H value. The predicted hairiness H value was compared with the actual hairiness H value. The results show that the BP network model has the highest accuracy of prediction hairiness H value, followed by the RBF network, while the multiple linear regression model has the worst prediction accuracy.
作者
陆奕辰
王蕾
潘如如
高卫东
LU Yichen;WANG Lei;PAN Ruru;GAO Weidong(School of Textile and Clothing, Jiangnan University, Wuxi 214122, China;Key Laboratory of Eco-textiles,Ministry of Education, Jiangnan University, Wuxi 214122, China)
出处
《东华大学学报（自然科学版）》
CAS
北大核心
2019年第5期682-687,共6页
Journal of Donghua University(Natural Science)
基金
中央高校基本科研业务费专项资金资助项目(JUSRP51631A,JUSRP11805)
江苏省研究生科研与实践创新计划资助项目(SJCX 17_0481)
