摘要
针对车刀表面粗糙度大、磨损区域图像难以准确分割的问题,提出一种基于均值漂移和自适应阈值分割的刀具磨损检测方法。这一方法通过频率域滤波过滤图像中无效的高频噪声信息,对降噪图像进行形态学处理。利用均值漂移聚类算法进行预分割处理,简化纹理信息,并突出磨损区域特征。通过自适应阈值分割算法准确提取刀具磨损区域,同时计算刀具的磨损宽度均值。试验表明,采用所提出的刀具磨损检测方法,分割准确率达到97.96%,能够有效判别刀具的磨损状态。
Aiming at the problems that the surface roughness of turning tools is large and the image of the wear area is difficult to accurately segment,a tool wear detection method based on mean shift and adaptive threshold segmentation was proposed.This method filters the invalid high-frequency noise information in the image through frequency domain filtration,and performs morphological processing on the noise-reduced image.The mean shift clustering algorithm is used for pre-segmentation processing,which simplifies the texture information and highlights the characteristics of the wear area.An adaptive threshold segmentation algorithm is used to accurately extract the tool wear area,and at the same time the average tool wear width is calculated.Experiments show that the segmentation accuracy of the proposed tool wear detection method reaches 97.96%,which can effectively distinguish the wear status of the tool.
出处
《机械制造》
2021年第5期71-75,共5页
Machinery
基金
浙江省自然科学基金资助项目(编号:LQ20E050006)
宁波市自然科学基金资助项目(编号:2019A610158)
宁波市科技创新2025重大专项(编号:2018B10005)。
关键词
刀具
磨损
检测
均值漂移
阈值分割
Cutting Tool
Wear
Detection
Mean Shift
Threshold Segmentation
作者简介
奚昌泽(1995-),男,硕士研究生,主要研究方向为工业检测。
