摘要
在接触式螺纹测量的过程中,采集的表面原始轮廓信号伴随的随机噪声对后续计算螺纹参数值的影响非常大,需要进行滤波处理以便后续参数计算。本文采用模拟的梯形螺纹原始测量数据作为数据源,通过较为常见的四种滤波算法,即低通类滤波器的中值滤波、高斯滤波、平滑滤波以及时域类滤波器的小波滤波,阐述各算法原理及优缺点,以螺纹数据整体的均方根误差、螺纹数据边缘的均方根误差和信噪比作为滤波效果的评价指标,对比四种算法的拟合效果、保真性和细节处理效果,优选出最佳的滤波算法。
In the process of contact thread measurement,the random noise associated with the original contour signal of the collected surface has a great influence on the subsequent calculation of thread parameters,so filtering processing is needed for subsequent parameter calculation.In this paper,the original measured data of trapezoidal threads are used as data sources.Four common filtering algorithms,namely median filter,Gauss filter,smoothing filter and wavelet filter of time domain filter,are adopted.The principles and advantages and disadvantages of these algorithms are described.The root mean square error and signal-to-noise ratio of the edge of the ridge data are used as the evaluation index of the filtering effect.By comparing the fitting effect,fidelity and detail processing effect of the four algorithms,an optimal filtering algorithm is selected.
作者
倪高烽
程维明
曹振
王海涛
赵璐绮
NI Gaofeng;CHENG Weiming;CAO Zhen;WANG Haitao;ZHAO Luqi(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《智能计算机与应用》
2020年第5期213-217,223,共6页
Intelligent Computer and Applications
基金
上海工程技术大学研究生科研创新项目(上海大学探索研究基金资助)。
关键词
滤波算法
螺纹测量
接触式
filtering algorithm
thread measurement
contact type
