EDGE LOCALIZATION BY MULTIPLE OF GAUSSIANs THEORY

作     者：高峰 Lucas J. van Vliet 

作者机构：Xi'an Institute of Optics & Precision Mechanics Academia Sinica Xi'an 710068 Faculty of Applied Physics Delft University of Technology 2600 GA Delft The Netherlans 

出 版 物：《Journal of Electronics(China)》 (电子科学学刊（英文版）)

年 卷 期：1995年第12卷第4期

页      面：367-373页

学科分类：07[理学] 070104[理学-应用数学] 0701[理学-数学] 

主　　题：Edge location Multiple-of-Gaussians Pseudo-second-derivative filter Zero-crossing 

摘      要：Zero-crossing of a derivative of Gaussian filter is a well-known edge location criterion. Examples are the Laplacian, the second derivative in the gradient direction (SDGD) and the sum of the Laplacian and SDGD (PLUS). Derivative operators can easily be implemented by convoluting the primitive image with a derivative of a Gaussian. Gaussian filter displaces the equipotential of half height inwards for convex edge and outwards for concave edges. A Difference-of-Gaussian (DoG) filter is similar to the Laplacian-of-Gaussian but with opposite sign and causes a convex edge shift inwards. This paper introduces the Multiple-of-Gaussian niters to reduce curvature-based location error. Using a linear combination of N Gaussians(N2) with proper weights, the edge shifts can be reduced to 1/(2N-3) of the ones produced by a similar Laplacian-of-Gaussian filter.