A hybrid genetic algorithm-based edge detection method for SAR image
Man Wang, Yuan Shuyuan · 2005
In this paper, a new edge detection method for SAR image using a hybrid genetic algorithm (HGA) is proposed depending on a full study about the characteristics of SAR images. According to this method, firstly some new types of edges are defined, and then the edge detection is reduced to an optimization problem. Not only original image data, but also some local information of edge, such as the continuity, thickness and regional difference of edges are included to define a cost function. Therefore, by the global searching capability of genetic algorithm, more continuous and accurate edges can be detected than other traditional methods. Moreover, a local optimization operator is employed to speed up the convergence of algorithm. So the method presents a remarkably rapider speed than classical genetic algorithm, as well as better edges. The simulations results also demonstrate its efficiency.