Wavelet Analysis for Crack Detection: A Case Study with Db12 Wavelet and Thresholding

Figure 7(a) presents the wavelet coefficients obtained after applying a 3-level wavelet decomposition using the Db12 wavelet. It is evident that the low-frequency components predominantly consist of noise coefficients. In contrast, the high-frequency components exhibit clear signs of crack coefficients, characterized by their prominence and reduced noise levels. Following the decomposition, we implement a thresholding process on the wavelet coefficients, as depicted in Figure 7(b). This technique selectively preserves significant coefficients while effectively suppressing insignificant ones by setting them to zero.