摘要
基于中心极限定理提出了一个高斯滤波器逼近模型,并利用冲激响应不变法设计了数字高斯滤波器。通过级联方法,该滤波器的幅度传输特性偏差可控制在(-0.28%～+1.14%)以内,扩展为二维高斯滤波器,其三维幅度传输特性偏差可控制在(-0.42%～+1.12%)以内。该滤波器采用递归实现算法,算法简便,易于实现。在递归算法的基础上,引入边缘初始值的预测方法,有效抑制了递归算法的边缘效应。实验结果表明:该滤波器能够稳定可靠地提取三维表面参考面,在普通计算机上处理2000×2000点的三维采样数据仅需要7s,效率很高。
Based on the central limit theorem,an approximation model of Gaussian filter is put forward.Then,a digital Gaussian filter is designed by combination with impulse invariant transformation and this model.Through the cascaded way,the amplitude transmission deviation of this filter can be controlled within （-0.28%～＋1.14%）.When extended to a two-dimensional（2D） Gaussian filter,the deviation of three-dimensional（3D） transmission characteristic can be controlled within（-0.42%～＋1.12%）.Its recursive algorithm is so simple that it is implemented easily.A method of forecasting the initial value of the edge is proposed to restrain the end effect caused by this recursive algorithm.Experimental results show that this filter can extract the reference surface of 3D surface stably and reliably,once extraction of 2000×2000 3D sampled data on a general computer only consumes 7s,efficiency is demonstrated.
出处
《工具技术》
2011年第9期83-86,共4页
Tool Engineering
