A cellular Automata approach for noisy images edge detection under null boundary conditions

Abstract

Cellular automata (CA) refer to a simple and conventional method which performs parallel processing, thereby exhibiting better performance than serial processing in certain contexts, particularly in terms of reduced time complexity. Edge detection is widely used in image processing and numerous methods have been proposed for this purpose. However, most of existing methods are serial techniques and fail to take into consideration noise content of the image. In this paper, an edge detection technique was proposed for noisy images based on a four-neighborhood under Null boundary cellular automata (FNNBCA) for noise elimination and a two-dimensional twenty-five neighborhoods under Null Boundary cellular automata (TFNNBCA) for edge detection. This method considers linear CA rules under null boundary conditions only. Efficiency of the proposed method was further compared to those of existing methods, indicating much promising performance of the proposed method for binary images, so that all edges were well detected even on complex images. Finally, results of implementing the method in MATLAB are presented.