Efficient edge detection method by using focus of attention area

Abstract

Edge detection is the foundation of many computer vision systems because of its highly repetitive computational intensive nature there have been many research efforts to find ways to improve both speed and quality of edge maps. This paper describes a method of edge segmentation based on the composition of the split and merge method and the Canny edge operator. In order to speed-up the computationally intensive Canny operator, the proposed method is based on preliminary region segmentation and finding a focus of attention area. The focus of attention area is formed as strips around boundaries of regions. The Canny operator is navigated to the focus of attention area and then applied. The proposed method has been tested on nontextural images of the blocks world scenes, toy images and the images of outdoor scenes. The experimental results are given.