Abstract
Edge features can be used in the scene analysis and computer vision. In this paper, we present a hierarchical relaxation method for detecting edge features in images. The relaxation processes are performed on the pyramid, which is a hierarchical representation about the image. Using multiresolution information about the image, we can change the weighted coefficients by considering the processed results at high hierarchy as a measurement of the local edge features in the image, satisfactory detection can be obtained in the all parts of the image, and convergence speed of the relaxation processes can be improved greatly. This method has the advantage of detection of complete border of the objects in the image.
