Edge Extraction Method Based on Separability of Image Features

This paper proposes a new robust method for detecting step and ramp edges. In our method, an edge is defined not as a point where the in tens it,^ changes largely, but as a region boundary based on separability of image features which can be calculated by linear discriminant analysis. Based on this definition of an edge, the intensity of an edge is obtained from tlie separability, wliich depends only on tlie shape of an edge. This feature enables us to easily select tlie best t,hreshold value for the extraction of an edge, our method can be applied to color, and texture edge extraction. Experimental results have shown that this proposed niethod is powerful enough to extract weak edges.