《活塞包胶组件外观缺陷的机器视觉检测系统研究本科学位论文》由会员分享，可在线阅读，更多相关《活塞包胶组件外观缺陷的机器视觉检测系统研究本科学位论文（69页珍藏版）》请在金锄头文库上搜索。
1、摘要 硕士学位论文题 目 活塞包胶组件外观缺陷的机器视觉检测系统研究 活塞包胶组件外观缺陷的机器视觉检测系统研究活塞包胶组件是摩托车减震器中的关键零件。相对于传统的活塞而言，它具有耐摩擦性和自润滑性好，复合强度高，使用寿命长，配合精度高，加工便利等优点。摩托车产业在我国的兴盛使得活塞包胶组件的生产规模也随之扩大。但是迄今为止，我国尚未开发出检测活塞包胶组件的外观缺陷设备，企业仍然依靠人力目视和经验来检测活塞包胶组件，严重制约了产品质量和生产效率的提高。因此，开发具备高速度、高可靠性的活塞包胶组件自动检测设备势在必行。活塞包胶组件形状复杂、缺陷多样，激光检测技术、涡流检测技术等表面缺陷检测技术不
2、适用于活塞包胶组件的缺陷检测。而机器视觉技术具有非接触、高速、高灵活性、高可靠性、高稳定性和高自动化程度等优点，在工业检测中有明显的优势，广泛地应用在实际生产中。本文在借鉴国内外相关技术成果和研究文献的基础上，设计了活塞包胶组件外观缺陷检测的机器视觉系统。论文的主要研究内容包括以下方面：1） 结合课题要求，分析了活塞包胶组件存在的缺陷类型，提出了活塞包胶组件外观缺陷的机器视觉检测系统方案；2） 在分析机器视觉系统硬件构成的基础上，完成了相机、镜头、图像采集卡及光源的选型。同时，分析并比较了各种照明方式，为活塞包胶组件的缺陷检测设计了光学照明系统和图像自动获取系统；3） 通过实验，对图像滤波和图
3、像分割的多种经典算法的处理效果做了比较，并在此基础上提出了适用于本课题的图像分割方法：利用OTSU算法和固定阈值法相结合的算法来分割上端面图像，下端面图像的分割采用了自适应阈值处理的方法。侧面图像的分割则是结合了形态学操作提取边缘和自适应阈值分割算法。4） 针对活塞包胶组件的各种缺陷提出了模式分类识别算法：端面图像的缺陷识别以极坐标变换为基础，提取相应区域处理分割出目标，再以目标连通域的面积为依据判定缺陷；侧面图像结合了组件侧面轮廓分析和连通域特性分析来检测不同的胶套缺陷。并通过实验验证了上述算法的效果。5） 利用MIL8.0函数库进行软件编程和调试，完成了软件系统的开发，实现了初始化相机，采
4、集图像，处理图像，控制输出等功能。论文的主要创新性在于设计了针对活塞包胶组件的缺陷检测算法，并开发了基于机器视觉的活塞包胶组件缺陷检测的软件系统，为整个系统的实现铺平了道路，同时也可以为今后类似的零件检测提供了参考和借鉴。关键词：活塞包胶组件 机器视觉 缺陷检测 图像处理IIIAbstractStudy on Machine Vision System for Inspection of Piston-covering UnitMajor Precision Instrumentation & MechanologyGraduate Xia Xinyi Supervisor Su Zhenwe
5、iPiston-covering unit is a crucial part of motorcycles vibration absorber. Compared with traditional piston, it is characterized by stronger frictional resistance and self-lubrication, higher combined strength and fit accuracy, long service life, processing facilities, and so on. With the developmen
6、t of motorcycle industry, the production scale of piston-covering unit developed from manual stage to automation stage. However, so far, there has not been any auto-inspection system for piston-covering unit in our country. Defect inspection of piston-covering unit depends on peoples eyes and manual
7、 skill. This seriously constraints the improvement of products and productivity. Thus, it is imperative to create a automatic system in high speed and reliability for inspection of piston-covering unit.Piston-covering unit has a complex form factor with diversified defections. The common inspection
8、techniques, such as laser detecting, eddy detection, etc. are not suitable for the defects inspection of piston-covering unit. However, a machine vision is of the advantages such as contactless, high speed, great flexibility, high reliability, good stability, high automatization and so on. Its disti
9、nct superiority allows it to be widely used in large scale production. Based on literature review, we designed a defect inspection system for piston-covering unit. The thesis includes the following sections:1) Analyzed the types of the defects on piston-covering units, and designed an automatic insp
10、ection system based on machine vision.2) Camera, lens, frame grabber and light sources were choosed according to their characteristics. Compared the various lighting sources, and designed optical lighting system for image acquisition of piston-covering unit.3) After comparing the results of multiple
11、 classic arithmetics such as image filtering and image segmentation in digital image processing, we designed an algorithm for our system: for the image of piston top surface, we combined OSTU algorithm and fixed threshold method. For the image of side, the algorithm is combination of adaptive thresh
12、old algorithm and morphological operation. At last, we use adaptive threshold algorithm to process the image of piston bottom surface. 4) We studied the pattern recognition algorithms for various defects of piston covering unit. For the images of top surface and bottom surface, we developed an algor
13、ithm based on polar coordinates and blob analysis; for the image of unit side, we complete the dectection by analysing contour line and blob informations. Then the performance of the algorithms is proved by experiments.5) The software for image acquisition and image processing were achieved based on
14、 Mil 8.0 function library.The innovations of this paper are that an algrithm is designed to recognise the defects of piston-covering unit, and a software system is developed for piston-covering unit based on machine vision technique. Our work paved a way to the realization of the auto-inspection of
15、the piston-covering unit, and can provide reference for the inspection of similar parts in the future.Keywords: Piston-covering unit, Machine vision, Defect inspection, Image processingVII目录目录1绪论11.1课题研究的目的及意义11.2国内外研究现状21.2.1活塞检测技术的研究现状21.2.2表面缺陷检测技术的研究现状31.2.3机器视觉技术及其发展41.3论文的主要工作及内容安排62活塞包胶组件视觉检测系统设计82.1活塞包胶组件主要缺陷分析82.1.1活塞铸件缺陷82.1.2胶套缺陷92.2活塞包胶组件视觉检测系统方案设计92.3系统的硬件组成102.3.1照明系统设计102.3.2图像采集设备142.4本章小结183活塞包胶组件缺陷检测算法研究193.1活塞包胶组件的图像预处理193.1.1图像滤波193.1.2图像分割213.1.3本文采用的图像分割方法243.2活塞包胶组件端面图像缺陷检测273.2.1活塞包胶组件端面图像圆心的确定273.2.2端面图像的极坐标变换283.2
