Weld defect identification through 3D scanning and error distribution modelling supported by point-feature histogram

The inspection of welded structures requires special attention, as the quality of the product is influenced by numerous factors. Determining the suitability of welds is a complex task. The aim of this work is to develop a method that effectively supports this evaluation process. The study presents a modern evaluation procedure based on PFH data and its practical application in the inspection of weld quality. Reference and defective point clouds were created using image processing and CAD modeling software, and then aligned using the ICP algorithm. This serves as the basis for generating data sets representing model errors. The results show that the distributions of the variables obtained in this way describe the nature of typical welding defects well. Automated evaluation of these distributions can reduce the turnaround time of inspections, which contributes to increasing the efficiency of welding processes.