An Adaptive Algorithm Based on Image Segmentation

A new algorithm for adaptive threshold segmentation based on combining Fisher criterion with location optimization is proposed in this paper. Fisher criterion is taking as the fitness function of genetic algorithm (GA), and an adaptive method which is used to calculate crossover probability and mutation probability is presented. Meanwhile, we add a new local optimization operator that solves the disadvantages of poor astringency and premature occurrence in GA. Experimental results show that the algorithm achieves better performance on convergence and robustness, can efficiently segment the details and converge the optimal threshold.