Fine tuning the GALE edge detection method

Most vision systems require the use of image processing applications that are highly dependent on the efficiency of edge detection techniques. These techniques are commonly implemented by applying an edge enhancement method followed by a thresholding point process. Most of these techniques are based on convolution algorithms that have a time complexity of O(n²) when the picture has size n×n. In order to reduce this time complexity, an improved solution depends on the reduction of the problem space. Such a reduction was recently achieved by a new method, named GALE, which combines the random search mechanisms of genetic algorithms with linear time methods. In this paper, a refinement of the GALE method is accomplished by introducing an iterative process that selectively eliminates from the population of the genetic algorithm those pixels that were previously identified as part of an edge. Experimental results show the improved performance of this method