采用切削刃重构的刀具磨损视觉检测方法

Visual Detection of Tool Wear Through Cutting Edge Reconstruction

针对目前难以在复杂恶劣的油污粉尘环境中实现对刀具图像的高质量采集和刀具磨损视觉特征的高精检测,对磨损缺失刀刃这一类最为典型且危害最大的刀具磨损开展研究,提出一种采用切削刃重构的刀具磨损视觉检测方法。首先,在数控机床加工台一侧搭建集成了一套具有镜头保护与清洁功能的图像采集装置,用于在机定期自动采集刀具磨损图像;然后,将采集的图像经以太网传输至计算机图像处理系统,利用设计的切削刃重构法对刀具磨损缺失区域进行切削刃重构,以此得到完整刀具图像,进而利用图像差分,将重构后的刀具图像与磨损刀具图像相减,实现刀具磨损缺失区域的自动识别;最后,基于识别的磨损特征测量刀具磨损的评估指标参数值,并判断是否需要换刀。实验结果表明:所提检测方法具有较大优势,解决了油污粉尘机加环境下刀具磨损图像采集困难的难题和难以从图像中分割识别刀具磨损缺失特征的难题,实现了刀具磨损的视觉高精高效检测;与现有的刀具磨损视觉检测系统以及现有的Canny边缘检测法、自适应阈值法等6种图像分割方法相比,所提方法避免了拆卸刀具进行离线显微镜检测和模板匹配的烦琐过程,可进行在机自动检测,同时平均检测准确率至少提升20%。

Currently,it is difficult to realize high-quality collection of tool images and high-precision visual detection of tool wear in complex and harsh oil and dust machining environment.For this reason,this paper proposes a cutting edge reconstruction method for visual detection of blade missing due to the most typical and harmful tool wear.First,an image acquisition device with lens protection and cleaning functions is built and integrated on one side of the NC machine tool processing table,to automatically collect tool wear images on a regular basis on the machine.Then the collected images are transmitted to the computer image processing system via Ethernet for reconstruction of the cutting edge missing from wear with the proposed method so as to obtain the complete tool image.After that,the reconstructed tool image is subtracted from the worn tool image by image difference,so as to realize the automatic recognition of the area missing from tool wear.Finally,based on the identified wear characteristics,the parameter values of the tool wear evaluation indexes are measured to determine whether to change the tool.The experimental results show that the method proposed in this paper have great advantages:It tackles the difficulty in tool wear image acquisition and segmentation to visually recognize the characteristics of the area missing from tool wear in the oil and dust machining environment,and realizes the high-precision and high-efficiency visual recognition and detection of tool wear.Compared with the existing six image segmentation algorithms including the visual recognition and detection system for tool wear,the Canny edge detection method,and the adaptive threshold method,the method proposed helps avoid the tedious process of disassembling the tool for offline microscope detection and template matching,and improves the average detection accuracy by at least 20%.