闽值分割是贴片机视觉检测中元器件识别的第一步,它的优劣决定了一系列后续算法的准确性,进而影响到最终元件贴装的精度。本文首先分析了常用的阈值技术及其优缺点,然后结合表面贴装元器件图像的实际特点,提出了一种基于均谷加权的快速阈值分割算法,用于表面贴装元器件图像的分割。对几种贴装元件图像进行实验仿真的结果表明,该算法的分割效果和处理速度都优于常用的阈值分割算法,满足了贴片机视觉检测的要求。
Threshold segmentation takes the first step of component image recognition in vision detection of Surface Mounting Machine. Its performance affects the accuracy of latter algorithms and determines the precision of component final position. The common methods of threshold techniques were analyzed, and a fast threshold segmentation algorithm was proposed for component image segmentation based on weighted peak values. The experiments results demonstrate that the proposed algorithm's performance is better than other two algorithms and satisfies the requirement of vision detecting in SMC (surface mounting component) mounting.