Using morphological operations — Erosion based algorithm for edge detection

Abstract

Edge detection remains a challenging task in many applications even with the abundance of commonly used methods such as Sobel, Robert, Prewitt, Canny, and Cellular Automata. In this paper, a new method is proposed for edge detection using morphological operations and utilizing erosion processes to identify the edges in an image. In this work we propose to use morphological operators of disk shape to detect thin edges from binary images generated using Otsu's thresholding technique. Using several synthetic and real images from a variety of applications, we show that our proposed algorithm provides a robust method in terms of accuracy and computation time, making it suitable for real time processing. The results also show that this algorithm is capable of producing one-pixel-width continuous edges as well as accurate positioning while preserving minute image details. We also present comparison results between our proposed method and Sobel, Robert, Prewitt, Canny and Cellular Automata methods.