摘要
针对高分辨率工业印刷电路板PCB图像的直方图和边缘特征进行了统计分析,提出了一种基于直方图统计的双门限选择方法进行像素级边缘定位,然后根据亚像素原理把像素级边缘校正到亚像素精度的快速亚像素边缘检测算法。与插值、空间矩以及曲面拟合等传统亚像素边缘检测算法的对比实验结果表明,本文算法在检测精度上达到0.066个像素,满足检测精度的要求,并在计算速度上具有较大优势。
A fast sub-pixel edge detection method based on gray histogram and double threshold for the high resolution PCB image is proposed.The experimental results show that the positioning accuracy amounts to 0.066 pixels.This algorithm can accurately locate edge and compute more quickly than the traditional sub-pixel edge detection algorithms.The proposed algorithm can efficiently fulfill the requirements for high accuracy and real-time performance.
出处
《微计算机信息》
2010年第10期3-4,59,共3页
Control & Automation
关键词
机器视觉
PCB
亚像素
边缘检测
高分辨率
machine vision PCB sub-pixel boundary detection high resolution
