Abstract: Aiming at the shortages of complex computing and low efficiency in traditional circle mark location method, this paper combines the quickness of point Hough transformation and the precision of sub-pixel, presents circle sub-pixel detection algorithm based on point Hough transformation and Legendre moments. It uses the algorithm to detect round location mark in PCB vision detection. Experimental results show that the positioning accuracy amounts to 0.056 pixels, it has the characteristics of anti-noise, accuracy, rapidity and robustness, and it can satisfy the requirement of high precision and real-time in PCB visual detection.

Key words: Printed Circuit Board(PCB) vision detection, circle mark location, point Hough transformation, Legendre moment

摘要： 针对传统圆形标志定位方法存在运算复杂、效率低的不足，结合点Hough变换的快速性和亚像素细分的精确性，提出基于点Hough变换与Legendre矩的圆亚像素检测算法，对印刷电路板(PCB)视觉检测中的圆形标志定位进行检测。实验结果表明，该方法定位精度可达0.056像素，具有抗噪性、准确性、快速性和鲁棒性的特点，能满足PCB视觉检测中高精度和实时性的要求。

关键词: 印刷电路板视觉检测, 圆形标志定位, 点Hough变换, Legendre矩

CLC Number: 

• TP393