Subpixel Estimation of Circle Parameters Using Orthogonal Circular Detector

In machine vision inspection, the accuracy with which edge points are located determines the effectiveness of inspection. Subpixel level accuracy is one of the methods that can improve the accuracy of edge location. Applying subpixel accuracy to locate edge points on line features has been discussed for several decades, but little research has focused on the features of curves even though circles are often used as features in industrial parts. In this research, a new algorithm named the orthogonal circular detector (OCD), was developed. For the subpixel estimation of circular edge points, This technique consists of five 9×9 masks based on a truncated basis system set and represents a circular detecting area. Instead of all the image data in the circular detection area being examined, only those points on the periphery are utilized, so the computation time can be dramatically reduced. When a segment of a circular object is located using the OCD, two intersecting edge points can be estimated. Since the coordinates of the two edge points are not limited to integers, subpixel accuracy can be achieved. The application of the OCD to industrial inspection is also discussed in this paper.