Generation of fuzzy edge images using trapezoidal membership functions

Abstract

The fuzzy representation of the edges has been widely studied in different works. Generally, for each pixel, the authors use membership degrees linearly proportional to the magnitude of the gradient at that position of the image. This would be equivalent to using a triangular membership functions on the gradient magnitude. In this work we study the use of parametric functions in the transformation of the gradient images into fuzzy sets. The results obtained with different parametrical functions are studied using Baddeley's Delta Metric.