基于机器视觉的内螺纹参数检测系统设计

发布时间：2018-04-05 07:14

本文选题：机器视觉　切入点：非接触检测　出处：《合肥工业大学》2017年硕士论文

【摘要】：在工业生产与加工中,螺母的作用不言而喻,不同使用环境下,对螺母各参数的质量要求不同,因此根据不同要求,螺母内螺纹的检测方法与效率也不相同。本文针对目前内螺纹检测方法中存在的检测效率低、精度低,对工件接触性检测会造成损伤等问题,提出了一种基于机器视觉的内螺纹高效检测的方法,并设计了检测系统。根据待测螺母的特点,通过从侧面对螺母照明与成像,实现了内螺纹图像的快速采集,对称式的激光三角测量法,极大地降低了由于螺母偏心造成的检测误差;针对所采集到的图像特点,通过调整图像灰度化的权值系数,获得对比度得到增强的灰度化图像,同时降低金属散射对图像的影响;对高斯模板滤波后的图像进行亚像素级的边缘检测,对边缘点进行最小二乘法的多项式拟合,将拟合曲线作为内螺纹轮廓线,在螺纹轮廓线上定位内螺纹特征点,利用特征点的坐标计算出内螺纹的相关参数。本文在理论分析与图像算法研究的基础上,搭建了系统试验平台,对系统的稳定性和测量随机误差做了实验测试与分析。系统对螺纹内径的重复测量极限误差为12. 78μm,单个工件的图像采集与处理时间不超过200ms。
[Abstract]:In the industrial production and processing, the role of nut is self-evident, under different use environment, the quality requirements of each parameter of nut are different, therefore, according to the different requirements, the testing method and efficiency of nut inner thread are also different.Aiming at the problems of low detection efficiency, low precision and damage to workpiece contact, this paper presents a high efficiency detection method of internal thread based on machine vision, and designs a detection system.According to the characteristics of the nut to be tested, the fast acquisition of the inner thread image and the symmetrical laser triangulation method are realized by illuminating and imaging the nut from the side, which greatly reduces the detection error caused by the eccentricity of the nut.According to the characteristics of the collected image, the grayscale image is enhanced by adjusting the weight coefficient of the grayscale image, and the influence of metal scattering on the image is reduced.The edge of the image filtered by Gao Si template is detected at sub-pixel level, and the least square polynomial fitting of the edge point is carried out. The fitting curve is regarded as the contour line of internal thread, and the characteristic point of internal thread is located on the thread contour line.The relative parameters of internal thread are calculated by using the coordinates of feature points.On the basis of theoretical analysis and image algorithm research, a system test platform is built in this paper. The stability of the system and the measurement of random error are tested and analyzed experimentally.The limit error of repeated measurement of thread inner diameter is 12.At 78 渭 m, the time of image acquisition and processing for a single workpiece is not more than 200 ms.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG85;TP391.41

【相似文献】