Edge detection using evolutionary algorithms
Corinna Ng, Wing-Nin Leung, Fan Chun · 2003
Edge detection is an important step in vision systems and object recognition. Existing edge detection operators such as the gradient operator and the Laplacian operator are based on the assumption that edges in an image are step intensity edges, therefore the resulting edges are usually thick and fragmented. Finding true edges of an image is still a difficult task. Another problem with most of the existing operators is huge search space. Considering an image with 1024 pixels by 1024 pixels, the solution space is 2/sup 1024/spl times/1024/. Therefore, without optimization, the task for edge detection is time consuming and memory exhausting. The paper presents the results of an experiment which evaluate the performances of three different evolutionary algorithms on edge detection. The three evolutionary algorithms applied in this experiment are genetic algorithms, tabu search and, evolutionary tabu search algorithm.