SWARM OPTIMIZATION ALGORITHM FOR BOUNDARY DETECTION IN IMAGES

Abstract

Swarm Intelligence uses a set of particles that move through the given search space and gather local information. This limited memory of particles is then used for solving problems. In this paper, we present algorithms based on the swarm intelligence based technique - the firefly algorithm. To generate boundaries, extra particles are removed by applying an appropriate threshold mechanism. In ACO, ants are distributed throughout the image and the movement takes place on the basis of pheromones, heuristics and the probability value. In the proposed firefly algorithm for Boundary Detection, fireflies are distributed in an image and the movement towards the solution depends upon the brightness and attraction amongst the fireflies. These algorithms’ performance is evaluated using Berkeley Segmentation DataSet (BSDS) and is compared with those of existing boundary detection algorithms via BSDS benchmark.