基于双平行平面相机模型的钢板尺寸视觉测量

发布时间：2018-03-08 01:20

本文选题：单目视觉　切入点：双平面相机模型　出处：《光学精密工程》2016年04期 　论文类型：期刊论文

【摘要】：提出基于双平行平面相机模型的视觉测量方法,用于测量生产线上运动钢板的尺寸。该方法采用数据驱动的方式计算像点在标定平面上投影点的世界坐标;采用k近邻(k-NN)方法生成目标在标定平面上的无畸变投影图像,并建立投影图像与世界坐标系的直接关联。提出了双平行平面模型下相机光心位置标定算法,利用线结构光进行板材厚度测量;在无畸变的投影图像上利用钢板边缘间的平行和垂直性进行钢板边缘特征提取,通过边缘直线的世界坐标方程求取长宽尺寸。最后,给出了针对大尺寸钢板测量的多相机测量系统框架。提出的方法为单目视觉测量方法,相比于其他方法具有现场安装简单和标定工作量小的特点。通过图像分辨率为640×480的相机对尺寸为80mm×50mm×15mm的标准铝块进行了测量,结果显示:厚度测量误差为0.1mm,长度和宽度的误差在0.2mm以内。实际应用中测量精度远高于加工精度,能够满足产品计量的要求。
[Abstract]:A visual measurement method based on the model of double parallel plane camera is proposed to measure the size of moving steel plate on the production line. The method uses data driven method to calculate the world coordinates of the projection points of the image points on the calibrated plane. The k-nearest neighbor k-NN) method is used to generate the distortionless projection image of the target on the calibration plane, and the direct correlation between the projection image and the world coordinate system is established. The thickness of plate is measured by linear structure light, and the edge feature of steel plate is extracted by parallelism and perpendicularity between the edges of the plate in the projection image without distortion. The length and width of the plate are obtained by the world coordinate equation of the edge line. Finally, The frame of multi-camera measuring system for large-size steel plate measurement is presented. The proposed method is monocular vision measurement method. Compared with other methods, the field installation is simple and the calibration work is less. The standard aluminum blocks of 80mm 脳 50mm 脳 15mm are measured by the camera with image resolution of 640 脳 480. The results show that the error of thickness measurement is 0.1mm, the error of length and width is less than 0.2mm. the accuracy of measurement is much higher than that of machining in practical application, which can meet the requirements of product measurement.
【作者单位】：沈阳理工大学信息科学与工程学院;
【基金】：国家留学基金资助项目(No.201308210280) 沈阳理工大学计算机应用重点学科开放基金资助项目(No.4771004kkfx10)
【分类号】：TP391.41

【相似文献】