A 2-Dimensional Continuous Wavelet Transform Technique for Composite Panel Inspections by Scanning Laser Doppler Vibrometry

Lamb waves propagation analysis is among the leading-edge approaches for the Non-Destructive Testing of composite thin-walled structures. The underlying diagnostic principle relies on the dependence of elastic waves wavelength on the plate thickness through which they propagate. Therefore, critical defects in composite panels, such as delaminations, can be detected by a non-contact, automated and high-resolution method such as Laser Doppler Vibrometry. The general approach applied in signal processing is referred to as Local Wavenumber Estimation. It consists of creating an image of the average wavelength over the acquisition frequency band in which the pixel element corresponds to a point of the vibrometer scan grid. The standard signal processing approach is typically implemented via the Short Space Fourier Transform. While effective at creating an average wavelength image over the acquisition frequency band, a major drawback of the Short Space Fourier Transform is its heavy reliance on the a priori selection of a spatial window size; a suboptimal choice can severely degrade detection robustness and spatial resolution. To overcome this limitation, we propose a novel method based on the 2D Continuous Wavelet Transform for the detection of defects in composite structures based on the isotropic Morlet wavelet. Thereafter, the proposed method is experimentally validated exploiting a glass fiber reinforced plate with an induced delamination. The proposed method proves superior on average by eliminating parameter dependency while preserving the original image resolution, resulting in a 66% Intersection over Union metric, comparable with the values of the standard method, but with a lower computational time.