机械零件视觉检测技术研究-机械设计及理论专业论文.docxVIP

摘 摘 要 1 摘 要 本文在研究近年来计算机视觉检测技术的发展状况基础上，结合 制造业信息化技术的发展要求，进行了机械零件的视觉检测技术分析 和研究，针对图像的噪声滤波、边缘检测、特征提取等问题进行了详 细的讨论，并研究了实际检测过程所必需的摄像机标定和调整技术。 在对图像进行分析的过程中，本文以圆盘形零件为例研究了 Prewitt算子、Sobel算子、Robert算子、Laplace算子等不同边缘检测 的提取效果，研究了一种自适应确定阈值的边缘跟踪检测算法，在一 定程度上提高了检测的精度。对于图像的特征提取，根据常用的Hough 变换，针对直线、圆、椭圆数字图像的特征，讨论了基本的特征检测 方法，并得出一种具有较高精度和检测速度的Hough变换的简化检测 算法，同时还对零件检测的面积、周长、圆形度等简单的形状特征提 取进行了讨论和研究。在摄像机的标定和调整上，通过对摄像机成像 的物理模型和圆形物体在摄像机中成像的几何模型，分析了摄像机标 定的方法，并根据圆投影特性提出了～种对摄像机定位调整方法，从 一定程度上减少摄像机畸变，提高了检测精度。 (通过以上工作，系统分析了机械零件视觉检测的关键技术，并自 行设计开发了一套视觉处理软件。F 关键词：机械零件 计算机视觉 图像处理边缘检测Hough变换 摄像机调整 摘 摘 要 Ⅲ The computer vision inspection system for the machine part is studied in the thesis based on the study of technology of computer vision inspection in recent years．This study can content the development of the manufactural information technology．The thesis firstly analyzes the selection of image sensor,frame—grabber board and light source．Further we particularly discuss the noise filtering，edge detection and feature extraction．And in the thesis，the camera calibration and its adjustment are also discussed and studied for the computer vision inspection． For the image process，the study is illustrated with a disc work part detection experiments relating to the method development and analysis． The thesis studies the results of Prewitt operator,Sobel operator,Roberts operator,Laplace operator’etc．and according to the defect of the Sobel operator introduced a self-adapting decision of threshold and edge tracking arithmetic．For the feature extraction，based on the basic shape feature inspection in the work piece inspection，the thesis discusses the feature extraction of the straight line，the cimle and ellipse．For detection of the straight line，the thesis introduces an extended Hough transform，which reduces the complexity of the parameter space and improves the operation speed．For the circle and ellipse detection，the thesis presents a method based on symmetry．At last，the thesis analyses the physical model of tampa