Automated dimensional quality assessment of steel members using TLS and enhanced edge line estimation

Dimensional quality assessment (DQA) of manufactured structural members is important to ensure that the final dimensions of steel components fall within acceptable manufacturing tolerances. Although a few studies have investigated the dimensional inspection of steel members, most rely on controlled or static experimental setups such as fixed cameras, stable lighting, or production-line systems. These constraints limit their applicability in on-site or flexible inspection scenarios. Thus, this study proposes a method leveraging terrestrial laser scan data that assesses two types of dimensional quality of steel members including overall length and cross-section dimensions. A new edge point detection algorithm and a new edge loss compensation model are both developed to accurately estimate the edges and corners of steel members. Experimental validations on various steel members with different sizes and cross-sectional shapes are conducted and results show that the proposed approach achieves an average error of 0.63 mm for overall measurements of a long member of H-beam and approximately 1 mm for cross-sectional measurements. Moreover, comparative analysis with previous algorithms proves that the method outperforms the prior edge estimation methods, demonstrating its potential for practical application.