摘要
固结磨料研磨垫的表面形态与其加工性能有着密切关系,为更好地了解固结磨料研磨垫表面形态,尤其是研磨垫中的金刚石、孔隙、金刚石脱落坑等的分布特征,提出一种基于深度学习的固结磨料研磨垫表面形态分析方法。首先,利用徕卡DVM6数字显微镜及其配套软件获取固结磨料研磨垫表面图像;然后,采用python3+OpenCV对图像进行预处理,并利用标注软件Labelme对图像进行标注,用于后续的训练和测试;最后,运用深度学习框架Tensorflow搭建Mask R-CNN模型。结果表明:Mask R-CNN模型能对单一固结磨料垫表面图像中的多目标进行有效分割与识别,其主要评价指标平均准确率达到78.9%,达到了图像识别的主流水平。
The surface morphology of fixed abrasive(FA)lapping pad is closely related to its processing performance.In order to understand the surface morphology of the FA lapping pad better,particularly diamonds,pores,and pits res-ulting from diamond falling off,a deep learning-based method for characterizing its surface morphology was proposed.First,the Leica DVM6 digital microscope and its supporting software were adopted to obtain the surface images of the FA lapping pad;then python3+OpenCV were chosen to preprocess the images,and the labeling software Labelme was used to label the images for subsequent training and testing data set;finally,the Mask R-CNN model was built using the deep learning framework Tensorflow.The results show that the Mask R-CNN model can effectively segment and recog-nize multiple targets in the surface image of a single fixed abrasive pad,and the average accuracy of the main evalu-ation indicators reaches 78.9%,reaching the mainstream level of image recognition.
作者
胡伟栋
王占奎
董彦辉
张召
朱永伟
HU Weidong;WANG Zhankui;Dong Yanhui;ZHANG Zhao;ZHU Yongwei(Mechanical and Electrical College,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;Mechanical and Electrical College,Henan Institute of Science and Technology,Xinxiang 453003,Henan,China)
出处
《金刚石与磨料磨具工程》
CAS
北大核心
2022年第2期186-192,共7页
Diamond & Abrasives Engineering
基金
国家自然科学基金联合基金(U20A20293)。
关键词
固结磨料研磨垫
深度学习
目标检测
图像处理
fixed abrasive lapping pad
deep learning
target detection
image processing
