Study Of Improved Ant Colony Algorithm Applied To Image Edge Detection

Ant colony algorithm is a combinatorial optimization algorithm which is formed according to the foraging rule of ants, originally people used it to solve the traveling salesman problem (TSP). Good view of ant colony algorithm to solve the second optimization problem, more and more scholars use ant colony algorithm to solve combinatorial optimization problems in various fields, including path planning, scheduling, network routing, business planning, image recognition and so on. The ability of individual ant is limited, but the entire colony was able to complete the impossible task for a single individual. This mechanism of self-organization is come from the pheromone. Ant colony algorithm is also simulated such a mechanism, so that ants search path according to the concentration of pheromone. However, this mechanism easily lead to local optimum in the colony, and a large number of ants will also make the computational speed of algorithm was slow. In order to solve these two problems, many researchers have proposed the improved methods.In this paper, the idea of genetic algorithm is used to improve the ant colony algorithm. The introduction of variability factors in the algorithm and the algorithm can be adjusted with the variation of factor values. So ant colony algorithm ensure the early stages of the search range is large enough to effectively prevent local convergence algorithm, while in the latter part of the algorithm to ensure that ants on pheromone sensitivity to accelerate the convergence speed. In this paper, the improved algorithm for a simulation experiment comparing, the final results show that the algorithm improved the speed of convergence, iterations, etc. are better than the basic algorithm to achieve the improvement.Image edge detection is an important element of the feature extraction, it belongs to the important areas of image segmentation. The aim is to find the pixels which have significant changes in the image and draw the overall outline of the image. Image Edge contains a wealth of information which reflects major differences between the target and the background, is the important basis to distinguish the target and the background. Edge detection methods typically are:gradient operator method, Sobel operator method, Robet operator method, Log operator method, Canny operator method. There are the presence of noise-sensitive situation in different levels. The ant colony algorithm for image edge detection, first create a matrix with the same size of the image. When the ants move in the image,they will leave pheromone along the way. For the pixels which changed significantly, more ants will visit them. By changing the pheromone to update mechanism, so that the ants tend to view the pixels which have high gray gradient value in the image. This paper uses an improved ant colony algorithm to realize the image edge detection.Experiments and simulation show that the algorithm is able to depict the image edge information more clearly, especially for the major edge.Changing the size of threshold can effectively inhibit the noise of the image.