A new type of effective morphologic edge detectors

Abstract

A type of edge detector based on the concept of gray-scale morphology is proposed. Edge detection can be divided into two phases; the first is noise removal, and the second is ideal edge detection. By using an iterative averaged closing-opening operation, impulse noise as well as Gaussian noise is eliminated from the image. Then, the resulting ideal edges can be extracted, either by using one of the classical spatial operators or by a simple morphologic operator. Results obtained from test images show that the proposed morphologic edge detectors are very effective.<>