摘要
提出了一种基于计算机视觉的轴承表面缺陷检测的新方法,采用A102fCCD数字摄像头作为图像传感器,利用数字图像处理技术对轴承表面缺陷进行检测,其主要步骤包括:图像的采集;采用边缘保持滤波器进行滤波、降噪;采用阈值法将图像二值化,并进行边缘检测;采用差影方法对有缺陷表面进行识别。试验证明,用该方法进行轴承表面缺陷检测是可行的,并具有方法简单、定位准确、抗噪能力强、计算速度快、实时性好等优点。
In order to perform non-contact inspection on surface defect of bearing, a new inspection method based on Charge Coupled Device （CCD） image, which use the A102fCCD digital camera as image sensor and utilizes image-processing technology, was proposed to inspect the surface defect of bearing. In the inspection system, the main approaches include collecting image, filtering image and reducing noise with the edge retention filter, creating a binary image with threshold method, detecting edge of the image, using the subtracting between two images to inspect surface defect. The experiment indicates that the above system has the advantages of convenience, correct location, the capability of noise resistance, high speed processing and temporal correctness.
出处
《机床与液压》
北大核心
2009年第9期130-132,共3页
Machine Tool & Hydraulics
基金
吉林省教育厅科技计划项目
东北电力大学博士基金资助项目(BSJXM-200501)
作者简介
作者简介：陈向伟（1960-），男，吉林省吉林市人，教授，博士。主要研究方向为计算机视觉检测、CIMS等。电话：13664329005，E—mail：xiangweichen@163．com。
