A Novel Subpixel Detection Method Based on Improved Hampel Circle Fitting for Ball Grid Array Components

Utilizing visual techniques for the accurate and reliable identification of ball grid array (BGA) components that have tightly packed and intricate pin configurations in surface mount technology (SMT) presents a significant challenge. We present a novel visual detection method for accurately measuring and aligning solder balls, enhancing the accuracy and stability of BGA recognition. The method involves extracting key dimensions and shape features of solder balls using the connected region analysis and geometric screening methods. We also provide a new interpolation method for estimating subpixel edge positions and an improved Hampel method for weighting solder ball edges to improve circle fitting accuracy. Furthermore, we propose a new approach for measuring the pitch and subsequently detecting balls in the X- and Y-directions. Experiments on SMT hardware setup demonstrate that our method offers superior measurement accuracy and stability compared to the existing methods.