Multiple-filtering process and its application in edge detection

Abstract

In this paper, a procedure of edge detection for a high dynamic range image with damaged edge information is proposed. This procedure is based on a scheme of multiple filtering processes which does not include any segmentation of the image. Three different filtering processes are designed to generate three gradient maps, in each of which gradients are calculated and modulated by using a specific filter. The enhanced gradients, i.e. those modulated correctly, are identified in each of the three gradient maps by using a selection algorithm. They are taken to generate a complete edge map. This procedure allows varieties of edge gradient enhancements applied in the same image by employing a set of simple filters without segmentation. The effectiveness of the detection process has been confirmed by simulations.