Related Experiment Video
Updated: Aug 7, 2025
11:34
Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography
Published on: May 15, 2017
11.2K
3D Scanner-Based Identification of Welding Defects-Clustering the Results of Point Cloud Alignment
János Hegedűs-Kuti1, József Szőlősi1, Dániel Varga1
1Faculty of Informatics, Savaria Institute of Technology, Eotvos Lorand University, H-9700 Szombathely, Hungary.
Sensors (Basel, Switzerland)
|March 11, 2023
Summary
This study presents a 3D scanner data framework for detecting welding defects. The method effectively identifies and classifies five out of six common welding deviations using density-based clustering.
Area of Science:
- Materials Science and Engineering
- Quality Control and Inspection
Background:
- Automated defect detection is crucial for maintaining weld quality.
- Traditional inspection methods can be time-consuming and subjective.
- 3D scanning offers a precise way to capture geometric data of welds.
Purpose of the Study:
- To develop and evaluate a framework for automated welding error detection using 3D scanner data.
- To classify detected deviations according to ISO 5817:2014 welding standards.
- To assess the effectiveness of density-based clustering for identifying and grouping weld defects.
Main Methods:
- Utilized 3D scanner data to create point cloud representations of welds.
- Employed density-based clustering algorithms to compare point clouds and identify deviations.
- Classified identified clusters into standard welding fault categories.
- Evaluated performance against six welding deviations specified in ISO 5817:2014.
Main Results:
- The framework successfully detected five out of six evaluated welding deviations.
- Errors were effectively identified and grouped based on point cluster locations.
- The method demonstrated robust defect localization and classification capabilities.
- A limitation was the inability to distinctly separate crack-related defects.
Conclusions:
- The proposed framework shows significant potential for automated welding quality control.
- Density-based clustering is a viable approach for identifying and categorizing geometric weld defects.
- Further refinement is needed to address the detection of specific defect types like cracks.
