基于像素分布特征的图像去噪法在刀具磨损检测中的应用  

Application of image denoising method based on pixel-distribution on tool wear detection

作　　者：廖玉松[1] 韩江[2] 

机构地区：[1]滁州职业技术学院机械与汽车工程系,安徽滁州239000 [2]合肥工业大学机械与汽车工程学院,安徽合肥230003

出　　处：《机械设计》2015年第6期93-98,共6页Journal of Machine Design

基　　金：国家自然科学基金资助项目(51275147);2011年安徽省省级质量工程项目--机械设计与制造省级特色专业建设(皖教高[2011]5号文件第121项);"高档数控机床与基础制造装备"科技重大专项:"标准型数控系统的产业化及专用型齿轮机床数控系统的研究开发"(2012ZX04001-021)

摘　　要：根据对机床刀具实时数据采集得到的图像的固有特征,提出了一种基于灰度差及像素分布连续性判断的方法,对磨损刀具灰度图像中的椒盐颗粒噪声采取先检测判断再进行去除的原则进行处理。采用Canny算子对图像的磨损区域进行边缘检测,检测结果的二值化图像中含有较多被算子误判为边缘的像素点。根据伪边缘像素点是分布特征,将其判断为噪声点,同样采用基于像素分布连续性判断的方法对伪边缘噪声点采取先检测后去除的方法进行处理,最终检测到磨损区域连续清晰的边缘线。结果证明:该方法在刀具磨损检测中有较强的实用性,在实践中可运用于刀具磨损的实时监测。A new image denoising method based on gray difference and pixels continuity was presented based on the inherent features of the real-time pictures from cutting tools. Firstly, noises were detected and then removed. Then, Canny operation was used to detect the edge of the tool wear area which made mistakes while detecting edges and then the fake-edge pixels were taken as noises in the binary image.And then, the method of pixels continuity was used again to detect the fake-edge and then remove them. Finally, the edge of the tool wear area was detected perfectly. The result shows that the image processing method is practical and can be used to monitor tool wear widely in engineering.

关 键 词：刀具磨损 噪声 灰度 边缘检测 二值化 

分 类 号：TB22[天文地球—大地测量学与测量工程]