An Approach for Accurate Edging using Dynamic Membership Functions

Abstract

In this paper, by means of fuzzy approaches, an accurate method is introduced for edging of color photographs. The difference between our method with other similar methods is the use of a morphological operation to think or thick the obtained edges. In this proposed method, a 3×3 window is dragged on the photo. For each block, 12 point sets will be defined, each including two non-overlapping point sets. Then, a fuzzy membership function will be designed for each point sets according to data of contrast. At last, the range of membership contrast degree for the points of second point sets will be assessed. A comparison of membership degree of second point sets with a pre-defined threshold indicates whether central point of window is bordered or non-bordered. The method was performed on some reference images and the results were compared with common edging methods. The results show that proposed method has a high capability to edge photographs