摘要
为了克服现有逆向工程装备价格昂贵且不能满足实时测量的缺点,提出了一种新型成本低、扫描速度快的三维激光自由曲面扫描系统。针对该系统的曲面重构问题,提出神经网络曲面重构方案,网络的输入选取所获得的点云数据的X、Y坐标,网络的输出则选取点云数据的Z坐标。比较了径向基神经网络(RBFNN)和多层前馈神经网络(MLPNN)两种典型神经网络曲面重构方案的优缺点。实验结果表明:RBFNN对于离散点云的曲面重构精度比MLPNN重构的精度高,但RBFNN较MLPNN所需的隐层神经元个数多。
The existed reverse engineering equipments are expensive and of low speed,a novel 3D laser free surface scanning system was developed,which had fast scanning speed and of low cost.As for the surface reconstruction part in the proposed system,two surface reconstruction schemes were presented based on two types of artificial neural network,namely,radial basis function neural network（RBFNN） and multiperceptron layer neural network（MLPNN）.The X and Y coordinates of point cloud data were chosen as network input while the Z coordinate was chosen as network output.Comparison is made between these two approaches.Experimental results demonstrate that the accuracy of RBFNN for scattered point cloud surface reconstruction is higher than MLPNN.However,more neurons in hidden layer are needed for RBFNN than MLPNN.
出处
《红外与激光工程》
EI
CSCD
北大核心
2011年第5期931-934,共4页
Infrared and Laser Engineering
基金
国家自然科学基金(50575029)
作者简介
王文标（1982-），男，博士生，主要从事物联网及视觉测量方面的研究。Email：WWb201@163．com
导师：马孜（1955-），女，教授，博士生导师，博士，主要从事数字化装备方面的研究。Email：zima@dlmu．edu．cn
