Ultrasonic NDT has been a well know approach to investigate material’s microstructures and structure integrity. However, the interpretation of ultrasound signals is difficult in composites materials because the phenomenon of the backscattering noise to hinder the real defect’s signal. The objective of this paper was to develop a new method to enhance defect detection. A denoising method by thresholding in wavelet domain is applied. The principle of this method is to replace the wavelet coefficients below a certain threshold by zeros and keeping or shrinking the remainders unchanged. In this work, we choose the Generalized Cross Validation (GCV) function to deal with the threshold minimization criterion case. The proposed method does not request any prior information about power of noise affecting the ultrasonic signals and has been tested on multiple real ultrasonic echoes coming from different materials. Finally, this approach can significantly reduce the noise while preserving the resolution of the original signal even though the signal to noise ratio is very low.

超声波无损检测一直是研究材料微观结构和结构完整性的众所周知的方法。然而，在复合材料中超声信号的解释是困难的，因为背向散射噪声的现象阻碍了真实缺陷的信号。本文的目的是开发一种新方法来增强缺陷检测。应用了小波域阈值去噪方法。该方法的原理是将低于某个阈值的小波系数用零代替，并保持或缩小余数不变。在这项工作中，我们选择广义交叉验证（GCV）函数来处理阈值最小化标准情况。所提出的方法不需要任何关于影响超声信号的噪声功率的先验信息，并且已经在来自不同材料的多个真实超声回波上进行了测试。最后，即使信噪比非常低，这种方法也可以显着降低噪声，同时保持原始信号的分辨率。