论文部分内容阅读
莫尔光栅的纳米级测量需要对莫尔信号进行高倍细分,而高倍细分的精度往往受到高斯白噪声的影响。将莫尔信号视为稳态模型进行去噪分析与处理时存在信号频率相对固定的缺陷,根据信号的频率是大范围可变的,且噪声分布在整个频率范围内,提出了一种更符合实际的时变模型,并采用小波阈值去噪法对信号进行处理。对时变莫尔信号进行了建模,对小波去噪原理及阈值去噪法进行了分析,经大量实验对比,选用Sym8小波基、分解尺度为6、阈值准则为Heursure的软阈值法去噪效果最好。去噪后,光栅莫尔信号接近理想信号,使莫尔信号的细分倍数达到1000倍。
The nanoscale measurement of moiré grating requires high-resolution subdivision of moiré signals, while the accuracy of high-resolution subdivisions is often affected by white Gaussian noise. The moir signal is considered as a steady-state model for denoising analysis and processing, there is a relatively fixed frequency of the signal defects, according to the signal frequency is a wide range of variable, and the noise distribution in the entire frequency range, a more consistent The actual time-varying model, and using wavelet threshold denoising method to signal processing. The time-varying Moiré signal is modeled, the wavelet denoising principle and the threshold denoising method are analyzed. After a large number of experiments and contrasts, the Sym8 wavelet basis is selected, the decomposition scale is 6, and the threshold criterion is the Heursure soft-thresholding method best effect. After de-noising, the grating Moire signal approaches the ideal signal, making the Moore signal’s subdivision multiple reach 1000 times.