This paper presents a multilevel edge detection technique using a novel improvement of particle swarm optimization algorithm, which computes the best thresholds that segment the image, while considering the spatial information of the pixel contained in such image. Using Kapur’s entropy as the optimized objective function, we introduced appropriate mechanisms to PSO aiming at speeding up the process of finding a solution, escaping the local optima, and enhancing accuracy. Then, edge detection is easily performed based on the thresholds provided by such a solution. According to our numerical experiments, our method exhibits good performance and segmentation accuracy.
Volume 11 | 03-Special Issue
Pages: 1756-1765