【作者】 王强；

【导师】 刘宝华； 高峰；

【作者基本信息】 燕山大学 ， 工程硕士（专业学位）， 2021， 硕士

【摘要】 汽车左前门车窗组合开关是汽车左前门门板的重要零部件,其出厂检测环节在汽车生产线中不可缺少。国内汽车零部件生产厂商对组合开关的检测都是由工人手动进行的,需分别拨动开关的5个按键,根据反馈的电信号判断是否正常。而组合开关每批次的检测数量都是以万计量的,工人检测任务繁重,且容易做出错误操作。为解决上述问题,设计了一种用于自动拨动组合开关按键的检测装置。本文主要对该装置的定位系统进行研究,不同车型的组合开关的按键分布位置不同,将机器视觉作为定位手段,可使装置在检测不同车型的组合开关时能够自动获取开关按键拨动点的位置,由于同一车型开关的按键结构和尺寸相同,在连续检测同一车型的组合开关时对该批次第一只组合开关进行定位即可。具体内容如下:首先,建立定位系统成像模型,通过左右图像内组合开关按键拨动点的像素坐标获取其在检测装置内的实际空间坐标。其次,为获取按键拨动点的像素坐标,将定位系统视觉检测任务确定为开关按键图标检测和升降键轮廓检测两个任务。针对各按键图标检测,提出基于轮廓几何特征检测和基于HOG特征检测两种方案,经干扰实验对比发现,基于HOG特征检测图标时受光线干扰影响较弱;针对升降键轮廓检测,提出基于Sobel边缘提取、HSV空间分割和Hough水平线搜索算法的升降键轮廓检测方法。再次,分析各种分类方式的适用性,最终采用两种分类器级联的方式来实现图标分类,可根据图标类别判断其对应的按键拨动点的属性。针对放大尺寸检测时图标模糊造成的分类不精确问题,提出基于高通滤波的二次锐化方法来强化图标分类。最后,根据定位流程,设计定位系统内PC端与控制器的通信协议,PC端与数据库的信息交互规则。综上,论文研究了装置定位系统的视觉定位方式,可实现对组合开关按键的自动定位,同时设计了定位系统信息交互方式,可实现定位数据的存储和传递,对实现汽车零部件自动检测具有重要意义。更多还原

【Abstract】 The combination switch of the left front door window is an important part of the left front door panel of the car and its inspection is indispensable in the car production line.In domestic auto parts manufacturers,the detection of combined switches is carried out manually by workers.The five keys of the switch need to be moved respectively to determine whether they are normal or not according to the feedback electrical signals.The detection quantity of each batch of the combination switch is measured in tens of thousands,and the detection task of workers is heavy,and it is easy to make wrong operation.In order to solve the above problems,a detection device for automatic toggle combination switch keys is designed.This paper mainly studies the positioning system of the device.Due to the different distribution positions of the keys of the combined switches of different models,the machine vision is used as a positioning method to enable the device to automatically obtain the position of the toggle point of the switch keys when detecting the combined switches of different models.Because the key structure and size of the switch of the same model are the same,the first combination switch of the batch can be positioned when the combination switch of the same model is detected continuously.The specific research contents are as follows:Firstly,the imaging model of the positioning system was established,and the actual spatial coordinates of the switch button toggle points in the detection device were obtained by combining the pixel coordinates of the left and right images.Secondly,in order to obtain pixel coordinates of key toggle points,the visual detection task of the positioning system was determined as two tasks: switch button icon detection and lifting key contour detection.For each key icon detection,two methods based on contour geometric feature detection and HOG feature detection are proposed.Through interference experiment comparison,it is found that the influence of light interference is weak when the icon is detected based on HOG feature.Aiming at lifting key profile detection,a lifting key profile detection method based on Sobel edge extraction,HSV space segmentation and Hough horizontal line search algorithm was proposed.Again,after analyzing the applicability of various classification methods,two classifier cascading methods are adopted to realize icon classification,and the attributes of the corresponding button toggle points can be judged according to the categories of icons.Aiming at the problem of inaccurate classification caused by blurred icons in magnifying size detection,a secondary sharpening method based on high-pass filtering is proposed to enhance the classification of icons.Finally,according to the positioning process,the communication protocol between the controller and PC in the positioning system is designed,and the information interaction rules between PC and database are designed.In summary,the paper studies the visual positioning mode of the device positioning system,which can realize the automatic positioning of the combination switch keys.At the same time,the paper designs the information interaction mode of the positioning system,which can realize the storage and application of the positioning data,which is of great significance to the realization of auto parts automatic detection.更多还原