多吸嘴贴片机的飞行视觉检测算法研究与系统开发

发布时间：2019-06-15 14:44

【摘要】：机器视觉系统是高性能贴片机的关键技术之一,在提高贴片机的速度和精度上起到十分重要的作用。针对多吸嘴贴片机元件和PCB基准点的快速采集与检测,本文提出基于飞行视觉的多吸嘴贴片机元件同时采集与检测算法和快速精确的基准点定位算法,有利于提高贴片机视觉检测的效率。主要的研究工作概括如下：1、深入调研贴片机及其视觉系统国内外研究现状和发展趋势,了解国内外贴片机整机研发的差距,分析高速多吸嘴贴片机及其视觉系统研究的难点和意义。2、针对传统贴片机飞行视觉只能同时处理少量元件的问题,提出一次性多元件图像同时采集的方法。对实现该方法的贴片机飞行视觉系统进行方案设计,确定所需的工业摄像机、镜头、光源等硬件,可获取吸嘴上多元件清晰的局部扫描图像；并设计面向该视觉系统的平面标定板和标定方法,实现摄像机的快速精确标定。3、为加快图像处理的速度,提出二值形态学和图像约化相结合的贴片元件和PCB基准点边缘区域自动提取方法。针对基于二维图像亚像素边缘提取算法复杂的缺点,提出先粗定位获得像素级边缘,再在此基础上利用一维曲线拟合法快速提取亚像素边缘的方法,以减少提取准确边缘的时间。4、针对贴片元件种类繁多的特点,将元件分成无引脚、球形引脚、伸出引脚三类,以提高检测算法的通用性和灵活性。针对无引脚元件类,提出一种基于直线海森范式和Tukey权重函数的最小二乘矩形拟合算法对多个元件亚像素边缘进行拟合,以降低因边缘不平整而产生的离群值对拟合精度的影响,快速精确检测出多个元件的中心位置和偏转角度。考虑PCB圆形基准点在成像过程中会变成椭圆形的特点,提出一种基于代数距离误差法和Tukey权重函数的最小二乘椭圆拟合算法对基准点亚像素边缘进行拟合,以减少因边缘污损而产生的离群值的影响,提高拟合精度和减少拟合时间,快速精确定位圆形基准点的中心位置。5、以龙门式精密运动控制平台作为硬件实验平台,运用HALCON和C++语言混合编程的方法开发多吸嘴贴片机飞行视觉系统,并进行实验验证和分析。
[Abstract]:Machine vision system is one of the key technologies of high performance placement machine, which plays a very important role in improving the speed and accuracy of placement machine. Aiming at the rapid acquisition and detection of multi-sucker placement machine elements and PCB reference points, this paper proposes a simultaneous acquisition and detection algorithm of multi-suction nozzle placement machine elements based on flight vision and a fast and accurate reference point location algorithm, which is beneficial to improve the efficiency of visual detection of placement machine. The main research work is summarized as follows: 1. The research status and development trend of the placement machine and its vision system at home and abroad are deeply investigated, the gap between the research and development of the whole machine at home and abroad is understood, and the difficulties and significance of the research of the high-speed multi-nozzle placement machine and its vision system are analyzed. 2, aiming at the problem that the flight vision of the traditional placement machine can only deal with a small number of components at the same time, the method of simultaneous collection of disposable multi-element images is put forward. The flight vision system of the placement machine is designed, and the required hardware such as industrial camera, lens, light source and so on can be determined, and the clear local scanning image of the multi-element on the suction nozzle can be obtained. The plane calibration board and calibration method for the visual system are designed to realize the fast and accurate calibration of the camera. 3. In order to speed up the image processing, an automatic extraction method of patch element and PCB reference point edge region combining binary morphology and image reduction is proposed. Aiming at the complexity of sub-pixel edge extraction algorithm based on two-dimensional image, a pixel-level edge extraction method based on rough location is proposed, and then one-dimensional curve fitting method is used to quickly extract sub-pixel edge in order to reduce the time of extracting accurate edge. 4. According to the variety of patch elements, the components are divided into three categories: no pin, spherical pin and protruding pin, in order to improve the universality and flexibility of the detection algorithm. A least square rectangular fitting algorithm based on straight line Hessen paradigm and Tukey weight function is proposed to fit the sub-pixel edges of multiple elements in order to reduce the influence of outlier values on fitting accuracy due to uneven edges, and to detect the center position and deflection angle of multiple components quickly and accurately. Considering the characteristic that the circular reference point of PCB will become oval in the imaging process, a least square elliptical fitting algorithm based on algebra distance error method and Tukey weight function is proposed to fit the subpixel edge of the reference point in order to reduce the influence of outliers caused by edge fouling, improve the fitting accuracy and reduce the fitting time, and locate the center position of the circular reference point quickly and accurately. Taking the gantry precision motion control platform as the hardware experimental platform, the flight vision system of the multi-nozzle placement machine is developed by using the mixed programming method of HALCON and C language, and the experimental verification and analysis are carried out.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN41

【相似文献】