International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 04 | Apr -2017
p-ISSN: 2395-0072
www.irjet.net
Image Edge Detection Using Modified Ant Colony Optimization Technique Shubhangi1, Shubham Singh2, Shubham Bisht3 1, 2, 3 UG
Scholar, Dept. of CSE, GCET, Greater Noida, UP, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract – Image edge detection is a technique of
segmenting an image into regions of discontinuity thereby marking sharp changes in intensity. Ant Colony Optimization is a meta-heuristic approach based on the foraging behavior of ants. This paper proposes a modified ACO algorithm to detect the edges in an image by updating the pheromone matrix twice and using weighted heuristics. The technique can be considered as an improvement to the original ant system. The approach presented in this paper can handle broken edges in an image and renders higher efficiency than other conventional techniques for edge detection. Key Words: Edge Detection, Pheromone Matrix, Ant Colony Optimization, Weighted Heuristics, Otsu Thresholding.
1. INTRODUCTION Image Processing is any operation that takes an image as an input and produces an output which maybe an image or a set of parameters related to an image. One of the most popular and complex problems of image processing is edge detection. An edge represents the contour features of the objects in an image and occurs as a sharp change in intensity from one pixel to another in an image. Image edge detection is a crucial part of image analysis and computer vision. It is a pre-processing stage in a number of applications in the areas of feature extraction. Various techniques have been proposed until now to detect the edges in an image. Some of these include Sobel, Prewitt, Laplacian and Canny operators. Ant Colony Optimization is another approach inspired by the natural behavior of ants while searching for food. The conventional approaches to edge detection involve higher computational costs as these operations are conducted for every pixel of the image. ACO is a probabilistic method that aims to find the optimized solution of the edge detection problem through a guided search over the solution space by constructing the pheromone matrix. In an ACO algorithm, ants move through a search space, the graph, which consists of nodes and edges. The movement of the ants is controlled by the transition probabilities, which reflects the likelihood that an ant will move from a given node to another. This value is influenced by the heuristic information and the pheromone information.
values are used and updated during the search. The algorithm is based on the fact that if two or more ants follow the same path, the Pheromone values are increased. The path along which the Pheromone value is maximum is taken as the shortest path resulting in an edge in the image. The performance of the ACO algorithm depends on route construction and Pheromone updates. The algorithm consists of three main steps. The first is the initialization process. The second is the iterative construction-and-update process, where the goal is to construct the final pheromone matrix. The construction-andupdate process is performed several times, once per iteration. The final step is the decision process, where the edges are identified based on the final pheromone values. In this paper, certain modifications have been made to the existing ACO approach which has led to higher quality of edges detected in an image, which is essential for different object recognition applications.
2. ANT COLONY OPTIMIZATION In this section, a theoretical discussion on the ant colony optimization meta heuristic and ant colony system, one of the main extensions to AS has been provided which describes in detail about extracting the edge information from the image using ACO. ACO is based on the food foraging behaviour exhibited by ant societies. Ants as individuals are unsophisticated living beings. Thus, in nature, an individual ant is unable to communicate or effectively hunt for food, but as a group, they are intelligent enough to successfully find and collect food for their colony. This collective intelligent behaviour is an inspiration for this evolutionary technique (ACO algorithm). The adoption of the strategies of ants adds another dimension to the computational domain. The ants communicate using a chemical substance called pheromone. As an ant travels, it deposits a constant amount of pheromone that other ants can follow [2]. When searching for food, ants tend to follow trails of pheromones whose concentration is higher. The pheromone deposited by the ants is subject to evaporation. The general procedure followed in an ACO algorithm is shown in Figure 1.
The heuristic information depends on the instance of the problem and it can be determined initially. Pheromone Š 2017, IRJET
|
Impact Factor value: 5.181
|
ISO 9001:2008 Certified Journal
|
Page 78