Automated Vision Inspection System for Cylindrical Metallic Components

The gap in manufacturing and assembly, in the automobile industry, is bridged by quality inspectors, who ensure that components manufactured are in proper condition to build a reliable system. However, if inspection is carried out by humans, it could lead to erroneous classification of defected components as good or vice versa, resulting in compromising of quality and wastage of material, respectively. Automation of this process is possible through the use of a vision-based inspection system and in this work such a system is proposed for cylindrical metallic components. The system uses computer vision and image processing methodologies such as morphology, edge and contour detection techniques to evaluate the errors effectively. Most systems evaluate errors stage by stage, which increases the consumption of space and cost of installation. The present inspection systems do not incorporate inspection of components that require rotation and some errors in small components cannot be detected. To not only address these issues but to also ensure reliability and efficiency, the proposed system, was tested for 250 samples, and was found to have an accuracy of 97%, which is comparable to the present systems.