针对工程中常用的脉冲反射法超声检测中反射回波信号的特点,研究了基于小波分解的阈值去噪方法,通过理论分析、仿真以及实验研究得出,对于超声回波选取sym8小波基,采用软阈值法,运用基于史坦无偏似然估计的阈值选取规则,分解层数为4层时,可以有效地去除噪声干扰,突出缺陷信号,去噪性能优于硬阈值法。
Aiming at the characteristics of uhrasonic reflected signal detected by using pulse-echo method commonly applied in engineering, the threshold denoising method based on wavelet decomposition is studied. It is concluded by theory analyzing, simulation and experimental research that ultrasonic reflected signal can be denoised effectively and the flaw echo can be signified by adopting soft-threshold method with sym8 wavelet and using the threshold rule of Stein's Unbiased Risk Estimate and 4 levels' decomposition. Also it is proved that the soft-threshold method is better than hardthreshold denoising method.