Real-Time Visual Quality Inspection System for Automotive Cable Manufacturing

In both society and industry, electric cables are used for various purposes. Companies often use different colors and materials to distinguish their properties. Quality control in cable manufacturing involves checking that colors meet specifications and that stripe widths (if present) are within the required range. Manual quality inspection methods are typically inefficient, time-consuming and error-proned, leading to uncaught defects. This paper introduces an automatic real-time visual inspection system that monitors the cable quality during its manufacturing process. The system employs image processing routines and four industrial Flir Blackfly S USB3 cameras to capture and analyze the entire cable perimeter in real-time. It evaluates the colors and thickness of the stripes to ensure they meet the predefined standards. The proposed solution aims to improve inspection accuracy and reduce the number of unnoticed defective cables that passes through the quality control. Experimental results show high precision and recall rates in segmentation, color verification and stripe evaluation tasks. The cable segmentation routine achieved a precision of 99.96%, while the stripe segmentation routine achieved 90.25%. The color verification routines for cables and stripes achieved precisions of 86.30% and 90.00%, respectively. The system stripe evaluation task achieved a precision rate of 79.74%. All image processing routines run in under 30 ms on a mid-tier performance workstation, demonstrating the system’s practical applicability and effectiveness in modern cable manufacturing lines.