Analytical Approach for Precise Localization and Quantitative Assessment of Surface Defects in Additive Manufacturing

Reliable assessment of surface defects has attracted considerable attention in the quality control of Additive Manufacturing (AM) products. Locating and assessing defect characteristics directly from surface topography is both inefficient and challenging. To address this, a method is proposed to localise defects by converting surface topography into a two-dimensional (2D) height map for training deep learning models. The world coordinates of the defect position are determined through the analysis of 2D images using deep learning and the correlation between these images and surface topography, enabling precise defect localisation and quantification within physical space. It is shown that this proposed method has reduced complexity, improved efficiency compared to existing methods and enabled quantitative analysis of defect features.