摘要
两圆柱正交相贯体相贯线的检测在工程领域具有广泛的应用,但有关相贯体相贯线的检测研究较少。提出了一种基于机器视觉的相贯线提取方法。在图像预处理阶段选取HSV空间的图像中S通道的图像进行OTSU二值化处理,并根据二值化图像进行了ROI图像分割。在边缘检测阶段,提出了一种基于频域图像增强的Canny边缘检测,将传入图像使用巴特沃斯高通滤波器进行图像增强,再经双边滤波后进行自适应双阈值的Canny边缘检测,最后利用闭运算图像形态学处理边缘图像,将图像中离断、孤立的边缘连接为整体。对于相贯线的拟合,分别采用概率霍夫变换和霍夫椭圆变换对直线、双曲线形的相贯线进行了提取。仿真模拟结果显示,所提方法相较于传统的Canny边缘检测方法可以更完整地检测出图像中的相贯线特征,相贯线的检测准确率达到96.2%,精度较高。
Detection of the intersection lines of two intersecting bodies with cylindrical shapes has wide applications in the engineering field.But there is limited research on the detection of intersection lines of intersecting bodies.A machine vision-based method for extracting intersection lines is proposed.In the image preprocessing stage,the image in the S channel of the HSV space is selected for OTSU binarization processing,and ROI image segmentation is performed based on the binarized image.In the edge detection stage,a frequency domain image enhancement-based Canny edge detection method is proposed.The incoming image is enhanced using a Butterworth high-pass filter,then subjected to adaptive dual-threshold Canny edge detection after bilateral filtering,and finally,the edge image is processed using morphological operations of closing to connect disjoint and isolated edges into a whole.For the fitting of intersection lines,the Probabilistic Hough Transform and Hough Ellipse Transform are adopted respectively to extract straight and hyperbolic intersection lines.Simulation results show that compared with traditional Canny edge detection methods,the method proposed can more completely detect the intersection line features in the image,and the detection accuracy of the intersection lines reaches 96.2%,demonstrating high precision.
作者
刘嘉伟
黄伟莉
胡积熠
Liu Jiawei;Huang Weili;Hu Jiyi(School of Mechanical and Electronic Engineering,East China University of Technology,Nanchang 330013,China)
出处
《机电工程技术》
2024年第5期178-182,共5页
Mechanical & Electrical Engineering Technology
基金
江西省新能源工艺及装备工程技术研究中心开放基金项目(JXNE2021-07)
2022东华理工大学实践教学类建设项目(DHSY-202212)。
关键词
相贯线检测
机器视觉
边缘检测
霍夫变换
intersection line detection
machine vision
edge detection
Hough transform
