Open Access

ARTICLE

A Multi-Level Threshold Method for Edge Detection and Segmentation Based on Entropy

Mohamed A. El-Sayed^{1, *}, Abdelmgeid A. Ali², Mohamed E. Hussien³, Hameda A. Sennary³

1 Department of Mathematics, Faculty of Science, Fayoum University, Fayoum, Egypt. Currently, Computer Science Department, Taif University, Taif, Saudi Arabia.
2 Computer Science Department, Faculty of Computers and Information Minia University, Minia, Egypt.
3 Department of Mathematics, Faculty of Science, Aswan University, Aswan, Egypt.

* Corresponding Author: Mohamed A. El-Sayed. Email: .

Computers, Materials & Continua 2020, 63(1), 1-16. https://doi.org/10.32604/cmc.2020.08444

Received 25 August 2019; Accepted 29 September 2019; Issue published 30 March 2020

Abstract

The essential tool in image processing, computer vision and machine vision is edge detection, especially in the fields of feature extraction and feature detection. Entropy is a basic area in information theory. The entropy, in image processing field has a role associated with image settings. As an initial step in image processing, the entropy is always used the image’s segmentation to determine the regions of image which is used to separate the background and objects in image. Image segmentation known as the process which divides the image into multiple regions or sets of pixels. Many applications have been development to enhance the image processing. This paper utilizes the Shannon entropy to achieve edge detection process and segmentation of the image. It introduces a new method of edge detection for 2-D histogram and Shannon entropy based on multilevel threshold. The method utilizes the gray value and the average gray value of the pixels to achieve the two dimensional histogram. The current method has apriority in comparison to some upper classical methods. The experimental results exhibited that the proposed method could capture a moderate quality and execution time better than other comparative methods, particularly in the largest images size. The proposed method offers good results when applied with images of different sizes from the civilization of ancient Egyptians.

---

Keywords

---

Citations