基于BP神经网络的结构光三维视觉检测方法研究

A Method of Structured Light based 3D Vision Inspection Using BP Neural Network

本文基于反向传播神经网络对结构光三维视觉检测方法进行了研究。首先介绍了结构光三维视觉检测原理 ,并对 BP算法的基本原理进行了阐述。在此基础上较详细地提出了用于神经网络训练和测试的样本点的空间坐标和图像坐标的样本数据获取方法以及光电瞄准装置的结构设计 ,最后利用样本数据对 BP神经网络进行了训练和测试 ,最佳 BP神经网络的训练精度为 0 .0 85m m,测试精度为 0 .13 2 mm。

Based on Back Propagation(BP)neural network,a method of structured light based 3 D vision inspection is proposed in this papre.The principle of structured light based 3 D vision inspection is firstly introduced,and BP neural network is also described.The method of gaining sample data,i.e.,frame coordinates and their image coordinates,is presented for the training and testing of BP neural network in detail.And an opto electrical aiming device is well designed.Finally,using the sample data,BP neural network is trained and tested.The training accuracy for the best BP neural network is 0.085mm,and its testing accuracy is 0.132mm.