刀具检测系统测头位姿的双目视觉检测研究

发布时间：2018-01-20 08:54

本文关键词： 双目视觉摄像机标定遗传算法角点提取极线约束位姿检测　出处：《西安工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：刀具检测系统的主要用途是在刀具设计、制造和重磨过程中对其几何参数进行测量,以保证刀具的生产质量;同时,在刀具用于生产过程中对刀具切削部分的几何形状及精度进行测量,用来保证刀具切削性能和产品质量。本课题采用双目立体视觉位姿检测方法实现对刀具检测系统扫描测头的位姿检测,以达到实现刀具非接触、在线测量目的。首先,基于计算机视觉基本理论,分析了双目视觉模型以及测量系统结构,详细阐述了双目视觉测量原理。其次,对摄像机标定方法及标定过程进行了详细分析,分析了标定过程涉及的图像坐标、摄像机坐标、世界坐标之间的转换关系以及摄像机畸变原因;重点研究了平面标定法的标定原理,并在此标定初始值的基础上用改进的遗传算法对摄像机标定,通过实验证明了该方法具有较高的标定精度。再次,对特征提取过程进行了研究,研究了特征提取原理,并分析了其常用的经典算法;在研究经典Harris角点提取方法的基础上,对其提取速度慢及像素级提取的缺点进行了改进,通过采用相邻像素取差以及与梯度灰度迭代相结合的方法,不仅提高了运算速度,而且实现了亚像素提取,并通过实验证明了该算法的有效性。最后,详细分析了位姿求解原理,通过极线约束进行立体匹配,运用视差原理求解测头空间位置,通过基于共面特征点求解测头空间姿态。利用搭建的双目视觉实验系统进行实验验证,通过实验数据结果对比分析,表明该方法够达到预期效果,完成位姿检测。
[Abstract]:The main purpose of the tool detection system is to measure the geometric parameters in the process of tool design, manufacture and regrinding to ensure the production quality of the tool. At the same time, the geometric shape and precision of cutting part are measured in the process of production. In order to ensure the cutting performance and product quality, the method of binocular stereo vision position and pose detection is used to detect the position and pose of the scanning probe of the tool detection system, so as to realize the tool non-contact. First of all, based on the basic theory of computer vision, the binocular vision model and measurement system structure are analyzed, and the principle of binocular vision measurement is described in detail. The camera calibration method and calibration process are analyzed in detail. The image coordinates, camera coordinates, the transformation relationship between the world coordinates and the camera distortion are analyzed. The calibration principle of the plane calibration method is studied emphatically, and the improved genetic algorithm is used to calibrate the camera based on the calibration initial value. The experimental results show that the method has high calibration accuracy. The process of feature extraction is studied, the principle of feature extraction is studied, and the classical algorithms are analyzed. On the basis of studying the classical Harris corner extraction method, the paper improves the slow extraction speed and the shortcomings of pixel level extraction. By adopting the method of adjacent pixel difference and combining with gradient grayscale iteration. It not only improves the operation speed, but also realizes sub-pixel extraction, and the validity of the algorithm is proved by experiments. Finally, the principle of position and pose solution is analyzed in detail, and stereo matching is carried out by pole line constraint. The spatial position of the probe is solved by the parallax principle, and the attitude of the probe is calculated based on the coplanar feature points. The experimental results are verified by the binocular vision experimental system, and the results of the experiment are compared and analyzed. The results show that this method can achieve the desired effect and complete the pose detection.
【学位授予单位】：西安工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG71;TP391.41

【相似文献】