Enhanced Identification of Internal Casting Defects in Vehicle Wheels Using YOLO Object Detection and X-Ray Inspection

Abstract

In the rapidly evolving automobile industry, the safety and quality of individual vehicle components have gained paramount importance. Among these, aluminum wheels are particularly critical, given their susceptibility to internal casting defects. This study presents a novel approach to identify these defects non-destructively, employing X-ray inspection and harnessing the power of YOLO (You Only Look Once) object detection. Images of vehicle aluminum wheels were obtained via X-ray inspection, revealing the presence of internal defects. Subsequently, a periodic noise e.limination algorithm, underpinned by morphological filtering and adaptive image processing weights, was utilized to enhance the image clarity. The application of a composite cascade filter further improved the image resolution. The enhanced images were then processed using YOLO object detection, a cutting-edge technology renowned for its precision in object detection tasks. This study explores the efficacy of different YOLO model architectures in detecting and identifying internal casting defects in aluminum wheels. Our research contributes to the development of a highly accurate system for the detection of internal casting defects in vehicle wheels, offering potential improvements in vehicle safety. This methodology, pairing X-ray inspection with advanced object detection algorithms, provides a robust approach for defect identification in the production process, laying the groundwork for future advancements in vehicle component quality control.