基于机器视觉的玻璃球损伤检测识别系统

发布时间：2018-04-15 21:11

本文选题：机器视觉 + Hough变换　；参考：《聊城大学》2017年硕士论文

【摘要】：玻璃球以化学稳定性好、机械强度高和电绝缘性好等特性广泛应用于众多领域。目前,国内玻璃球生产量和应用量巨大,然而玻璃球损伤检测技术落后,大都采用传统的人工方法检测;传统检测方法容易造成人工视觉疲劳、视觉损伤;效率低、误检率大,从而降低产品质量;急需一种自动检测玻璃球损伤的设备,有效解决人工检测的困境。本文提出了一种基于机器视觉的玻璃球检测识别系统;系统主要分为软件和硬件两个部分,软件部分主要是设计检测识别算法,获得玻璃球缺陷的信息;硬件部分主要完成图像的采集、物料传送和次品剔除工作。本文主要创新性的工作如下:1.玻璃球图像的采集。使用实验室MV-EM120C/M型号面阵CCD相机,环形LED台灯做光源高角度照明;提出了光源位置、照明亮度与相机拍摄角度及景深相结合的玻璃球图像采集方法;配以黑色绒布作背景,既增大被测物与背景对比度,又起到固定玻璃球防止滚动的作用。2.图像预处理。提出了基于小波变换的图像增强、基于hough变换平面圆识别、log算子边缘检测和数学形态学函数填充相结合的方法进行图像预处理,突出轮廓边缘、获取去背景的玻璃球图像。3.特征提取与缺陷检测。提出了基于区域标记的面积测量方法提取玻璃球图像缺陷特征、计算缺陷阈值,实现缺陷检测。4.制作GUI可视化操作系统界面。更直观的操作避免操作冗长程序,处理结果清晰明了,操作方式简单,易于人工交互。
[Abstract]:Glass spheres are widely used in many fields because of their good chemical stability, high mechanical strength and good electrical insulation.At present, the domestic glass ball production and application amount is huge, but the glass ball damage detection technology is backward, mostly uses the traditional manual method to detect; the traditional detection method is easy to cause artificial vision fatigue, the visual damage, the efficiency is low, the false detection rate is big,In order to reduce the quality of the product, it is urgent to use a kind of equipment to detect the damage of glass ball automatically, and to solve the problem of manual inspection effectively.In this paper, a glass ball detection and recognition system based on machine vision is proposed, which is mainly divided into two parts: software and hardware.The hardware part mainly completes the image collection, the material transfer and the defective product elimination work.The main innovative work of this paper is as follows: 1.The collection of glass ball images.Using laboratory MV-EM120C/M plane array CCD camera and ring LED lamp as high angle illumination, a method of glass ball image acquisition is proposed, which combines the position of light source, illumination brightness with camera shooting angle and depth of field, and is matched with black velvet as background.It not only increases the contrast between the measured object and the background, but also acts as a fixed glass ball to prevent rolling. 2.Image preprocessing.Image enhancement based on wavelet transform, edge detection based on hough transform plane circle recognition log operator and mathematical morphology function filling are proposed to pre-process the image, highlight the contour edge, and obtain the glass-sphere image. 3.Feature extraction and defect detection.An area measurement method based on area marking is proposed to extract the defect feature of glass ball image, calculate the defect threshold, and realize the defect detection. 4.Make GUI visual operating system interface.More intuitive operation to avoid lengthy procedures, processing results clear, simple operation, easy to manual interaction.
【学位授予单位】：聊城大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ171.65;TP391.41

【参考文献】