在小波变换轮廓术中,采用一维小波变换去解调变形结构条纹包含的物体面形分布时,小波处理所带来的误差分布同被测物体的面形分布及其变化率存在联系.为了减小小波变换轮廓术中相位提取的误差,从小波变换的相关实质出发,提出了利用神经网络的强大学习能力和函数逼近功能来校正相位.该方法以提取相位的一阶和二阶差分、小波尺度因子等相关计算量作为多层前馈型反向传播网络的输入,通过对复杂样本的训练,实现了输入与误差之间的非线性映射.计算机模拟和实验验证了用神经网络校正相位方法的有效性.
In WTP method, the phase information demodulated from the deformed fringe pattern is just the approximation of the real phase distribution caused by the height variation of the tested object. The error is related to the second derivative of the phase distribution and other parameters. In order to eliminate the measurement error of WTP, we employ the powerful ability of studying and function approximation of BP Neural network to minish the measurement inherent error of WTP. The computer simulation and a primary experiment verified our analysis.