Abstract: Image segmentation is the foundation of the image processing and analysis. The Otsu segmentation algorithm and genetic algorithm are analyzed in this paper, in order to improve the running performance of the algorithm, simulated annealing is introduced to put forward a kind of improved genetic simulated annealing algorithm (GASA). The whole running process of this algorithm was controlled by the temperature cooling schedule, with the improved Otsu method being used as the fitness function of the genetic algorithm. After several rounds of computing, an optimal threshold value was obtained for image segmentation. The experiments’ results showed that the image segmentation based on the GASA could be good at enhancing the comprehensive search ability of the algorithm, and avoiding the genetic algorithm’s falling into local optimization. Meantime, it would not only converge to the optimum segmentation threshold faster and more steadily, but also obtain higher segmentation quality.

Key words: image segmentation, GA, GASA, Otsu, threshold