A New Image Threshold Technique based on Metaheuristics

In a wide range of applications in image processing, the gray levels of pixels form a general representation for the image. In the meantime, thresholding which present the image into the binary state is a simple but effective tool for separating of the objects from the background. Image thresholding denotes a process by which an image is divided into two areas and each area is homogeneous. This paper proposes a new method of image thresholding using the optimal histogram thresholding by using Kapur's entropy method based on an imperialist competitive algorithm. In Kapur's approach, the main purpose is to maximize the entropy of gray levels distribution. The gradient descent is used to maximize the entropy in this method whereas using gradient descent enforces several suppositions which might not be easy to adjust in many situations. Experimental results show that using an imperialist competitive algorithm instead of gradient descent in Kapur's method makes a powerful system that is comparatively toward the ordinary algorithms.

Keywords:

Entropy-based thresholding , Imperialist Competitive Algorithm , Maximization