印制电路板图像对准算法研究

使用模板匹配法检测印刷电路板(PCB)质量,对标准图和待检测图之间的图像对准具有很高要求.在定位标志为圆的情况下,应用Hough变换虽然行之有效,但存在占用内存多,计算量大等问题.图像对准分为模糊对准和精确对准两个步骤,利用设置定位圆的先验知识实现模糊对准,并由模糊对准给出定位圆的相关参数,将这些参数运用于精确对准,能使图像的边缘提取和Hough变换变得简单、易行,减少了内存占用,提高了算法的处理速度.

Research of image alignment method for PCB

To detect the quality of Printed Circuit Board (PCB) by template matching needs accurate alignment of the standard image and the detected image. When the location mark is a circle, Hough transform can be used effectively to get the situation of the center, byt has defects in memory occupation and calculation. The image alignment was composed of fuzzy alignment with the prior knowledge of the circle and accurate alignment with circle location parameters given by fuzzy alignment. This method makes the edge extraction and Hough tranform simple and easy to realize, decreases the occupied memory and increased the processing speed.