A Sub-Pixel Localization Algorithm on the Basis of Gauss Curve Fitting in Gradient Direction
The most important thing of the detection system is the locating accuray of the pattern edge. With respect to the conventional sub-pixel localizaton algorithm and the irregular edge of the target image, a Gauss curve fitting edge localization algorithm in gradient direction is presented based on the Sobel operator and the usual curve fitting edge localization algorithm; according to a gradient distribution of a target image, a Gauss curve is used to fit defected edges so as to realize sub-pixel location of the edges. Experiments demonstrate that the proposed method is rightness and effectiveness, can obviously improve the accuracy of edge localization and target recognition.
Helen Zhang, Gang Shen and David Jin
N. Zhao and C. X. Fu, "A Sub-Pixel Localization Algorithm on the Basis of Gauss Curve Fitting in Gradient Direction", Advanced Materials Research, Vols. 219-220, pp. 1149-1152, 2011