Segment-based adaptive neighborhood orthogonal projection method for automatic complex concrete crack width measurement

• A fully-automated algorithm for complex concrete cracks width measurement is proposed. • Adjacent branch interference is reduced by a divide-and-conquer strategy. • Adaptive neighborhood orthogonal projection is adopted to eliminate manual threshold specification. • The proposed model demonstrated superior accuracy and robustness compared to conventional models. Accurate measurement of crack width in concrete structures was critical for structural performance assessment and the development of effective repair strategies. Existing methods faced significant limitations in handling complex crack morphologies, often resulting in measurement inaccuracies. To address this issue, a segment-based adaptive neighborhood orthogonal projection method was proposed for the quantitative analysis of concrete crack images, and the applicability of the method was evaluated using both real and synthetic cracks. The results indicated that the proposed method accurately identified both the location and magnitude of the maximum crack width in real concrete cracks, with errors of only 1.19% and 3.07% for meandering and complex reticulated cracks, respectively. Moreover, through the evaluation of synthetic cracks, the method demonstrated excellent accuracy and robustness, with maximum crack width measurement errors for multi-branch cracks, complex intersecting cracks, and ring-shaped intersecting cracks all remaining within ±2%. In addition, the proposed method outperformed conventional approaches in multi-point measurements on synthetic cracks, achieving a Pearson correlation coefficient of 0.97. These findings provided three main technical contributions: (1) the reduction of adjacent branch interference in complex cracks through a divide-and-conquer strategy, (2) employing an adaptive neighborhood orthogonal projection to eliminate the need for manual threshold specification, and (3) demonstrating strong adaptability of the proposed method to boundary irregularities. Together, these contributions offered an effective solution for quantifying and assessing the severity of complex cracks in concrete.