【作者】 肖乐萍； 王桂棠； 冯宝祥； 潘煜熙；

【Author】 Xiao Leping Wang Guitang Feng Baoxiang Pan Yuxi (School of Information Engineering,Guangdong University of Technology,Guangzhou 510006,China)

【机构】 广东工业大学信息工程学院；

【摘要】 论述一种利用线阵 CCD 作图像获取设备,用机器视觉完成型材尺寸非接触测量的方法。文中详细论述了通过图像获取设备的标准通信协议 TWAIN 驱动设备自动获取型材测量图像的关键步骤。基于型材的特点,在图像预处理中应用了迭代阈值化分割、均值滤波、罗伯特交叉算子等方法提取出清晰的型材图像边界。文中通过系统结构和工作过程分析了影响测量精度的因素。采用直线和曲线拟合的方法及随机 Hough 变换,来拟合型材边界,从而重构型材的几何图形,实现型材尺寸的精确测量。作者通过实验证明了这些方法的应用,使用4800DPI 的线阵 CCD 扫描仪对型材进行测量实验,与人工测量值对比,绝对误差不大于一个像素,精确度达0.05mm,能满足本实验中对型材的精度需求。更多还原

【Abstract】 A dimesional measurement methods of molding applying the machine vision techniques is presented.The measurement process is described,Where the TWAIN and the key steps are applied to let the image processing soft- ware to acquire the images automatically.This article introduces the image preprocessing,iterative threshold seg- mentation,median filtering and Robert crossorer operator.The methods of straight line fitting and curve fitting as well as the random Hough transfer are introduced.The measurement speed and accuracy of this system are im- proved by applying these methods.From the sides of systematic structure and working process,the influence factors of measuring precision are analyzed.更多还原