Vision-Based Real-Time Measurement System for Predrilled-Hole Inspection on Wooden Planks

Manual inspection of pre-drilled holes in wooden planks is labor-intensive, inconsistent, and prone to dimensional errors that can disrupt downstream assembly in furniture manufacturing. To address this gap, we present a low-cost, vision-based measurement system that automates hole detection and dimensional verification on white wooden planks. Using a consumer-grade USB camera housed in a custom enclosure, OpenCV image-processing pipelines, and calibrated coordinate transforms, the system processes live video streams to locate hole centers, and measure diameters in real time. We evaluated nine processing pipelines with combinations of Otsu, binary, adaptive thresholding, Canny, Sobel, Laplacian edge detectors that is under realistic lighting conditions. The optimal pipeline (binary thresholding + Laplacian edge detection) achieved a 95%-hole detection rate with zero false positives. While the system achieved sub-millimeter accuracy for centrally located features, deviations increased at the image periphery due to uncorrected distortions, highlighting the need for full geometric calibration. The prototype operates at 10–15 FPS on standard PC hardware, demonstrating a scalable, Industry 4.0-ready alternative to manual inspection.