Application of Machine Vision for Inspection of Drilled Components in Bolted Connections of Prefabricated Steel Structures

This paper presents a descriptive-applied review that examines the application of machine vision as an innovative approach for the geometric inspection of drilled plates used in bolted connections of prefabricated steel structures. Currently, in steel fabrication plants, the geometric characteristics of bolt holes including hole diameter, center location, center-to-center spacing, edge distance, and conformity with fabrication drawings are typically inspected by the quality control unit using manual instruments such as measuring tapes, calipers, gauges, and templates. However, these conventional methods are associated with several limitations, including being time-consuming, highly dependent on inspector expertise, and prone to human error, particularly under mass production conditions. To address or mitigate these limitations, this study proposes a practical and cost-effective framework for automated inspection based on machine vision and image processing techniques. The primary objective is to demonstrate the capability of a digital inspection framework that can enhance or partially replace manual inspection procedures. The proposed framework comprises image acquisition, camera calibration, image processing, edge detection, extraction of the geometric features of holes, and comparison with design specifications and allowable tolerances in order to identify common drilling defects or discrepancies such as diameter deviation, hole center misalignment, improper edge distance, and nonconforming hole patterns-prior to final assembly of the components. Furthermore, to improve accuracy, particularly in images with reduced quality due to unfavorable workshop lighting conditions, it is recommended to integrate lightweight deep learning models with classical image processing algorithms. Analytical findings indicate that machine vision can increase inspection speed, reduce human error, and establish a traceable image-based database of inspected components. Nevertheless, factors such as ambient lighting conditions, image quality, camera orientation, and the need for precise calibration remain significant challenges in the practical implementation of this approach.

کلیدواژه ها:

Machine vision ، image processing ، steel component inspection ، bolted connections ، drilling ، quality control