Ultrasonic Signal De-noising Using Dual Filtering Algorithm

This paper describes the proposal of dual filtering algorithm used for ultrasonic A-scan signal de-noising. In ultrasonic non-destructive testing the signals characterizing the material structure are evaluated. The sensitivity and resolution of typical ultrasonic systems is commonly limited by backscattering and electronic noise level commonly contained in acquired ultrasonic signals. In ultrasonic non-destructive testing area it is really difficult to detect flaw in materials with coarse-grain structure. In this paper, the main goal was to efficiently suppress the undesirable noise level and successfully detect the fault echo that is hidden under the noise level. The dual filtering algorithm, presented in this paper, is based on application of efficient denoising algorithms as Wiener filter and discrete wavelet transform. The Wiener filter based group delay statistics is presented and the best parameters are searched. The discrete wavelet transform algorithm with different mother wavelets, threshold levels and threshold rules is also presented. Both algorithms are evaluated using two parameters, signal-to-noise ratio and fault echo corruption. The proposed dual filtering algorithm is performed on both simulated and real ultrasonic signals acquired on grainy material used for constructing airplane engines.