Automatic inspection system of adhesive on vehicle windshield using computational vision
Polyurethane-based adhesives are applied on the windshields of vehicles in the automotive industry to fix the windshield and seal the cabin. A failure in the adhesive bead could allow water to ingress between the windshield and the vehicle body. If not detected in the leak test, it can lead to high...
| Autores: | , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2023 |
| País: | Brasil |
| Institución: | Universidade Estadual Paulista (UNESP) |
| Repositorio: | Repositório Institucional da UNESP |
| Idioma: | inglés |
| OAI Identifier: | oai:repositorio.unesp.br:11449/248314 |
| Acceso en línea: | http://dx.doi.org/10.1007/s40430-023-04051-x http://hdl.handle.net/11449/248314 |
| Access Level: | acceso abierto |
| Palabra clave: | Automatic bead inspection Computer vision Low-cost hardware Open-source software |
| Sumario: | Polyurethane-based adhesives are applied on the windshields of vehicles in the automotive industry to fix the windshield and seal the cabin. A failure in the adhesive bead could allow water to ingress between the windshield and the vehicle body. If not detected in the leak test, it can lead to high cost due to warranty repairs, inconvenience to customers and damage to the brand. Commercial solutions are available in the market to detect an interruption in the adhesive bead right after its application on the windshield, before it is fitted to the vehicle, but at high cost. This paper proposes an automatic inspection system based on computer vision, low-cost hardware, programming in Python language and making use of open-source libraries. A batch of defect-free windshields was inspected using the proposed inspection system. In the impossibility of obtaining defective parts for validation, windshield images were modified to simulate defects and the images were evaluated by the developed algorithm. The algorithm showed quite good results at the end, and we could establish the system's effectiveness at 100% for defect detection capability and 21% of false detections. |
|---|