Noise Filtering for MoireFringe Signals Based on Variable Step Size Adaptive Neural Network Algorithm

摘要

An adaptive filtering algorithm based on neural network is used to restrain noises of MoireFringe signals. How to realize this algorithm and its efficiency are studied followed. After analyzing the noises sources, the reason of adopting this method is discussed. The realization of the algorithm and how to adjust the step size are made out. We subdivide the signals filtered by way of tangent later. Based on comparing the characteristics of the original and the created signals, it is showed that the subdivision precision is improved. When we subdivide circular grating MoireFringe to 512 times by using this algorithm, the maximum error is 1.236" in one individual pulse. The effect of this algorithm is better than routine filtering methods because it has wide frequency range in improving the MoireFringe signals quality and excellent data preparation for subdivision.