A General Chip Subpixel Segmentation Localization Method Based on Improved Mayfly Algorithm

Abstract

In order to improve the robustness of processing chip images in surface mount technology, especially when it comes to using a single threshold value under different lighting conditions, this paper aims to propose a chip localization algorithm with low computational complexity and high generality. The study investigates the multi-threshold- based online chip localization problem and introduces an intelligent optimization algorithm to enhance its performance. An automatic adjustment mayfly method is presented, improving the mayfly algorithm by combining it with the sine and cosine algorithm to enhance global search and convergence capabilities, resulting in improved fitness values. Additionally, image processing using inter-class variance yields multiple thresholds. Together with corner point detection, a versatile chip localization method is proposed. Simulation results demonstrate significant enhancements in solution accuracy, convergence speed, and merit- seeking capability achieved by the improved algorithm. Finally, the method's effectiveness is verified through various chip localization experiments.