应用函数链神经网络测量管道内壁缺陷尺寸  

Application of Functional Link Neural Network in Size Measurement of Pipeline Inner Wall Flaws

作　　者：王庆峰[1] 张全法[1] 高晓鹏[1] 

机构地区：[1]郑州大学物理工程学院,河南省郑州市450001

出　　处：《微计算机信息》2012年第4期33-34,共2页Control & Automation

摘　　要：基于图像测量管道内壁缺陷尺寸时,图像上单位像素所代表的实际尺寸与很多因素有关,如摄像机的内部、外部参数以及摄像机到管壁的距离等,是一个复杂的非线性关系。在摄像机参数固定的条件下,采用函数链神经网络来逼近图像上单位像素所代表的实际尺寸与管道内径之间的非线性关系,进而实现了无需标定摄像机参数的管道内壁缺陷尺寸测量。实验结果表明,这种方法具有易于实现、误差小等特点。When measuring flaw size of pipeline inner wall based on image, there are many factors that influence the actual pixel size, such as camera＇s internal and external parameters, distance from camera to pipeline inner wall, ere, and the relation is a complex nonlinear function. On condition that camera parameters are constant, use functional link neural network to approximate the nonlinear relation between actual pixel size and pipeline internal diameter, which makes possible of size measurement of pipeline inner wall flaws without calibration of camera oarameters. Exoeriment results show that this method is easy to realize and with high accuracy.

关 键 词：管道内壁缺陷 缺陷尺寸测量 函数链神经网络 管道内径 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]