Influence of thresholding procedures in ultrasonic grain noise reduction using wavelets

Area of Science:

• Materials Science
• Signal Processing
• Non-Destructive Testing

Background:

• Ultrasonic non-destructive testing (NDT) is crucial for evaluating materials, especially those with strong sound scattering properties.
• Enhancing the signal-to-noise ratio (SNR) is vital for accurate flaw detection in ultrasonic NDT.
• Wavelet transform techniques offer a powerful approach for signal denoising in complex material evaluations.

Purpose of the Study:

• To investigate the impact of various wavelet transform parameters on SNR enhancement in ultrasonic NDT.
• To analyze different thresholding methods and threshold selection rules for optimal denoising performance.
• To evaluate the effectiveness of discrete wavelet transform (DWT) with decomposition level-dependent thresholds.

Main Methods:

• Utilized discrete wavelet transform (DWT) for signal denoising.
• Investigated various wavelet types, thresholding methods, and threshold selection rules.
• Employed synthetic grain noise with embedded flaw signals and experimental ultrasonic data from composite materials for evaluation.

Main Results:

• The choice of wavelet, thresholding technique, and selection rules significantly influences denoising performance.
• Demonstrated effective SNR enhancement using DWT with decomposition level-dependent thresholds.
• Validated the denoising approach on both synthetic and real-world ultrasonic data from carbon fibre reinforced plastic composites.

Conclusions:

• Wavelet transform-based denoising is effective for improving SNR in ultrasonic NDT of scattering materials.
• Optimizing processing parameters like wavelet type and thresholding is essential for robust flaw detection.
• The findings provide a basis for developing more accurate ultrasonic inspection techniques for advanced materials.