Purpose <p>This study aims to develop a noise-robust crack detection algorithm based on finite element modelling for a single cracked cantilever&#xa0;beam. The objective is to accurately identify the crack location from beam deflection data, even in the presence of measurement&#xa0;noise.</p> Methods <p>A numerical model simulates the cantilever beam with a single crack, introducing a local slope discontinuity in its elastic line. To&#xa0;detect this discontinuity obscured by noise, a discrete wavelet transform (DWT)-based algorithm is proposed. Experimental validation&#xa0;is carried out using an economical, digital photograph-based setup to capture beam deflection shapes. Beam edges are extracted&#xa0;from images, and the non-cracked edge profile is used to determine the deflection shape. The DWT-based technique is applied to&#xa0;this data to locate the crack. To improve noise robustness, a windowed variance technique is incorporated to amplify crack-induced&#xa0;discontinuities. Furthermore, averaging the DWT coefficients from multiple beam responses is employed to suppress noise and&#xa0;enhance crack detectability.</p> Results <p>The simulation and experimental results demonstrate that the proposed DWT-based algorithm effectively identifies the crack location,&#xa0;even in noisy measurement conditions. The use of windowed variance improves sensitivity to slope discontinuities, while averaging&#xa0;wavelet coefficients from multiple beam images significantly enhances detection performance.</p> Conclusions <p>The study concludes that the combined use of discrete wavelet transform, windowed variance, and coefficient averaging provides a&#xa0;reliable and noise-robust method for crack detection in cantilever beams. The proposed economical, photograph-based experimental&#xa0;approach further makes the method accessible for practical structural health monitoring applications.</p>