基于机器视觉的锻件尺寸检测方法  

Forgings Size Inspection Method Based on Machine Vision

作　　者：苗杰 倪洪启[1] 杨兵 MIAO Jie;NI Hongqi;YANG Bing(School of Mechanical and Power Engineering,Shenyang University of Chemical Technology,Shenyang 110142,China;Jiangsu Shengxian Forging Co.,Ltd.,Changzhou 213169,China)

机构地区：[1]沈阳化工大学机械与动力工程学院,沈阳110142 [2]江苏圣贤锻造有限责任公司,江苏常州213169

出　　处：《机械工程师》2025年第3期39-41,47,共4页Mechanical Engineer

摘　　要：针对锻件尺寸测量过程中的人工手动接触性检测速率低下及检测不方便等问题,设计了一种基于Halcon的锻件尺寸检测系统。拍摄照片之后,先对图像进行合适的预处理,将原始图像转化为灰度图,然后选取待测区域,采用高斯滤波以削弱高频噪声,接着使用双边滤波对图像进行处理,它考虑了像素之间的空间距离和像素之间的灰度差异,从而能够去除噪声并保留图像边缘的细节,然后进行图像增强处理。通过二值化阈值分割筛除噪声进而进行特征筛选,然后进行形态学处理腐蚀操作,最后对锻件尺寸进行测量并将数值显示到窗口上,对参照物区域进行分割,分别得到参照物的像素尺寸和实际尺寸,再通过数学运算分别得到900、1000、1100℃状态下待测锻件的实际尺寸。试验结果表明,该算法能够很好地对待测锻件进行处理,进而得到尺寸数据。To solve the problems such as low speed of manual contact detection and inconvenient detection in the process of forging size measurement,this paper designs a forging size detection system based on Halcon.After the photo is taken,the image is preprocessed properly,the original image is converted into grayscale image,and then the area to be measured is selected,Gaussian filtering is adopted to reduce high-frequency noise,and then the image is processed by bilateral filtering,which takes into account the spatial distance between pixels and the grayscale difference between pixels,so as to remove noise and retain the details of the image edge.After that,image enhancement processing is carried out,binarized threshold segmentation is performed to screen noise and then feature screening is performed,and morphological processing corrosion operation is carried out,and finally the forging size is measured and the value is displayed on the window,and then the reference area is segmented to obtain the pixel size and actual size of the reference respectively.The actual dimensions of the forgings to be measured at 900,1000 and 1100℃are obtained by mathematical calculation.The experimental results show that the algorithm can process the measured forgings well and obtain the dimensional data.

关 键 词：锻件测量 图像处理 双边滤波 二值化 特征提取 形态学处理 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]