基于计算机视觉的细丝直径精密测量方法研究

发布时间：2018-05-14 21:30

本文选题：畸变校正 + 亚像素　；参考：《西安理工大学》2016年硕士论文

【摘要】：随着测量技术的飞速发展,图像测量成了近年来测量领域中出现的新的技术。伴随着计算机技术、激光技术、微电子技术等新技术的发展和应用,以及测量技术本身的进步,图像测量为测量技术提供了很多新的方法和手段。在测量学中,测量误差与精度是对测量结果进行分析的重要指标,在基于图像的测量中,如何通过分析图像来校正测量误差,提高系统的测量精度,是图像测量技术中一个重要的研究方向。本文通过图像测量的方法,并引入亚像素边缘检测算法和相机畸变校正方法,以及利用自行设计的精密物体检测硬件平台,进行了细丝直径的精密测量方法的研究。通过对细丝图像特征的分析,应用了亚像素级边缘检测算法来检测细丝边界,提高边缘定位的精度,另外,通过设计标定网格模板定位点精确提取的方法,利用校正公式进行硬件检测平台中相机畸变的校正,具体的工作介绍如下:1、首先设计了一种使用远心镜头并搭配合适的相机和光源来进行细丝检测的成像系统,并实现了该硬件系统。为了获得清晰的成像,设计了对物距进行精密调节的方法。通过利用远心镜头放大倍率不变的特性,以及设计的网格标定模板和校准丝标定方法实现细丝直径的精密测量的系统。2、在使用平面圆形网格模板对相机畸变进行标定的过程中,选取三个标定点,为了获得标定点的精确坐标,以标定点区域的“十字形”交叉点作为形态特征进行坐标提取。在实现精确坐标提取的过程中,首先综合使用了多种图像预处理的算法,并提出了一种先利用网格边缘点求出拟合直线,用拟合直线交叉点得出像素级交叉点坐标,然后再利用空间矩亚像素算法获得“十字形”交叉点精确坐标的方法。3、为了获得相机的像素与细丝真实物理尺寸的标定关系,采用了三根已知宽度的光刻校准丝来对放大倍数进行标定,在测量校准丝宽度的过程中,使用基于多项式拟合的亚像素级边缘检测算法来获得高精度边缘坐标。最后,采用MATLAB作为编程语言,实现了细丝直径精密检测系统的界面及算法,并采用毛绒作为试验样品,实现了完整的测量过程,测量结果验证了系统的精度,并验证了系统设计的合理性。
[Abstract]:With the rapid development of measurement technology, image measurement has become a new technology in the field of measurement in recent years. With the development and application of computer technology, laser technology, microelectronics technology and measurement technology itself, image measurement provides many new methods and means for measurement technology. In the field of surveying, measurement error and precision are the important indexes to analyze the measurement results. In the image-based measurement, how to correct the measurement errors by analyzing the images to improve the measurement accuracy of the system, It is an important research direction in image measurement technology. In this paper, we introduce sub-pixel edge detection algorithm, camera distortion correction method and the precision object detection hardware platform designed by ourselves to study the precision measurement method of filament diameter through image measurement method and sub-pixel edge detection algorithm and camera distortion correction method. Based on the analysis of filaments' image features, a sub-pixel edge detection algorithm is applied to detect the edge of filaments to improve the accuracy of edge location. The correction formula is used to correct the distortion of the camera in the hardware detection platform. The specific work is as follows: 1. First of all, an imaging system using telecentric lens and matching the appropriate camera and light source to detect the filaments is designed. The hardware system is realized. In order to obtain clear imaging, a method of precise adjustment of object distance is designed. By using the invariant magnification of telecentric lens and the system of precision measurement of filament diameter using the grid calibration template and calibration wire calibration method, the camera distortion is calibrated by using the planar circular mesh template. In order to obtain the exact coordinates of the scalar points, the cross crossing points of the marked fixed points are used as the morphological features to extract the coordinates. In the process of exact coordinate extraction, various image preprocessing algorithms are used synthetically, and a fitting line is first obtained by using grid edge points, and pixel level intersection coordinates are obtained by fitting straight line intersection points. Then we use the spatial moment sub-pixel algorithm to obtain the precise coordinates of the "cross" intersection. In order to obtain the calibration relationship between the camera pixels and the real physical size of the filaments, Three lithographic calibration wires with known width are used to calibrate the magnification. In the process of measuring the width of calibrated wire, a sub-pixel edge detection algorithm based on polynomial fitting is used to obtain the high-precision edge coordinates. Finally, the interface and algorithm of the fine wire diameter precision detection system are realized by using MATLAB as programming language, and the plush is used as the test sample to realize the complete measurement process. The measurement results verify the accuracy of the system. The rationality of the system design is verified.
【学位授予单位】：西安理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TH74;TP391.41

【参考文献】