基于改进引导滤波算法的蜗轮齿廓边缘保持方法  

Maintain edge method of worm gear tooth profile based on improved guided filter algorithm

作　　者：郑永[1,3] 谢伟 陈艳 赵化雨 张天恒[1,3] 段高歆 刘宏荻 ZHENG Yong;XIE Wei;CHEN Yan;ZHAO Huayu;ZHANG Tianheng;DUAN Gaoxin;LIU Hongdi(Chongqing Key Laboratory of Time Barrier Sensing and Advanced Detection Technology,Chongqing University of Technology,Chongqing 400054,China;School of Mechanical Engineering,Chongqing University of Technology,Chongqing 400054,China;Engineering Research Center for Mechanical Detection Technology and Equipment,Ministry of Education,Chongqing University of Technology,Chongqing 400054,China;School of Electrical and Electronic Engineering,Chongqing University of Technology,Chongqing 400054,China)

机构地区：[1]重庆理工大学时栅传感及先进检测技术重庆市重点实验室,重庆400054 [2]重庆理工大学机械工程学院,重庆400054 [3]重庆理工大学机械检测技术与装备教育部工程研究中心,重庆400054 [4]重庆理工大学电气与电子工程学院,重庆400054

出　　处：《机电工程》2024年第12期2152-2162,共11页Journal of Mechanical & Electrical Engineering

基　　金：国家重点研发计划项目(2023YFB3209400);重庆市教育委员会科学技术研究项目(KJQN202301161,KJZDM202301104)。

摘　　要：在使用机器视觉方法测量蜗轮齿距过程中,工业相机采集的蜗轮齿廓图像极易受到高斯噪声和光照不均引起的轮廓伪影的影响。针对以上问题,提出了一种基于改进引导滤波算法的蜗轮齿廓边缘保持方法。首先,对原始蜗轮齿廓图像进行了滤波处理和二值化处理,滤波处理采用了改进的引导滤波算法,使用高斯函数构造权重系数替代了引导滤波算法中的均值权重系数,二值化处理使用最大类间方差进行了阈值分割;然后,对滤波后的图像与二值化图像进行了图像融合,对融合后的图像进行了拉普拉斯锐化处理;最后,进行了灰度拉伸对数变换,得到了增强后的图像,并采用实验验证了基于改进引导滤波算法的蜗轮齿廓边缘保持方法。研究结果表明:基于改进的引导滤波的蜗轮齿廓边缘保持算法在处理蜗轮齿廓边缘时,相比传统保边滤波算法,峰值信噪比可提高7.81%,图像对比度可提高43.6%,有效地增加了图像的信噪比和对比度;使用该蜗轮齿廓边缘保持算法增强后的蜗轮齿廓图像,当采用机器视觉测量其蜗轮齿距偏差时,左、右齿面单个齿距偏差为4.6μm和4.5μm,左、右齿面齿距累积偏差为16.5μm和16.6μm,测量结果均在允许的齿距偏差范围内。这说明采用蜗轮齿廓边缘保持算法增强后的蜗轮齿廓图像,去噪效果更好,蜗轮齿廓细节更清晰准确,有利于后续蜗轮齿廓特征的提取与齿距的测量。In the process of measuring worm gear pitch using machine vision methods,the worm gear tooth profile images captured by industrial cameras were highly susceptible to the influence of Gaussian noise and contour artifacts caused by uneven lighting.To address these issues,a worm gear tooth profile edge-preserving method based on an improved guided filtering algorithm was proposed.Firstly,the original worm gear tooth profile image was filtered and binarized.The filtering process used an improved guided filtering algorithm,constructing weight coefficients with a Gaussian function to replace the mean weight coefficients in the guided filtering algorithm.The binarization process was employed using the Otsu s method for threshold segmentation.Secondly,the filtered image was fused with the binarized image,and the fused image underwent Laplacian sharpening processing.Finally,a grayscale stretching logarithmic transformation was applied to obtain the enhanced image,and the worm gear tooth profile edge-preserving method based on an improved guided filtering algorithm was validated through experiments.The experimental results show that,in comparison to the traditional edge-preserving filtering algorithms,the worm gear tooth profile edge-preserving algorithm based on the improved guided filtering increases the peak signal-to-noise ratio by 7.81%and increases the image contrast by 43.6%,it effectively improves the signal-to-noise ratio and contrast of the image.The worm gear tooth contour images enhanced by the edge-preserving algorithm are utilized for machine vision measurement of the worm gear pitch deviation.The individual pitch deviations for the left and right tooth surfaces are respectively measured at 4.6μm and 4.5μm,while the cumulative pitch deviations for the left and right tooth surfaces are respectively 16.5μm and 16.6μm.All measurement results fall within the allowable range of pitch deviation.The research findings indicate that the worm gear tooth contour images enhanced by the edge-preserving algorithm exhib

关 键 词：蜗轮齿廓图像 引导滤波 二值化 图像融合 图像锐化 灰度拉伸 图像增强 

分 类 号：TH132.44[机械工程—机械制造及自动化]