A locating algorithm based on OGHT for PCB mark orientation

In this paper, we focus on the accurate orientation of printed circuit board (PCB). In PCB manufacturing, it is critical to do accurate localization during online registration procedure. Mark point can be used to adjust the positions and orientations of PCB. Generalized Hough Transform (GHT) is a robust algorithm for object recognition. However, it has some disadvantages in industrial real time application such as computational complexity, large memory. To improve efficiency, an optimized Generalized Hough Transform (OGHT) method, based on the direction of the gradient vectors and scale segmentation can increase computation speed and reduce needed memory space. Experimental results demonstrate that the speed of our proposed algorithm is approximately ten times faster than that of GHT and it is robust under various noisy conditions, illumination changes and blur conditions.