Machine vision-based detection method for buzzer iron cores.

Abstract

As a fundamental acoustic component, the buzzer is widely used in various electronic systems. The iron core is a critical element in buzzers for supporting the coil, and it is currently fed primarily by mechanical methods. To further improve the automatic feeding efficiency of iron cores, a machine vision-based detection method is proposed to achieve core localization, pose recognition, and notch-angle measurement. The method first employs the Hough transform to locate iron cores on the vibratory tray and exclude overlapping cores. It then statistically counts the edge pixels around each core's center to select only those facing upward. Finally, by traversing the core's circumference, it pinpoints the notch localization and computes its angle. This providing the necessary data support for the automatic grasping and placement of iron cores by the manipulator. Experimental results demonstrate that the Hough transform algorithm adopted in this paper achieves a mean relative localization error of only 2.61%, a recognition precision of 100% for front-up iron cores, and an average notch-angle measurement deviation of 1.06°. Compared with YOLOv8, the proposed method offers clear advantages in both detection accuracy and practicality.