Based on the principle of hormone modulation in endocrine system, an improved adaptive genetic algorithm was proposed. This algorithm designed an adaptive crossover operator and an adaptive mutation operator based on the downward form of Hill function, to make the crossover rate and mutation rate self-regulate according to the diversity of the population in each generation, and maintain the diversity at a reasonable level in the whole process of evolution. Experimental results of 4 test functions and 3D brain image segmentation showed the improved genetic algorithm can not only effectively maintain the diversity of the population, but also avoid premature, and performances of this algorithm are better than those of the other two adaptive genetic algorithms and traditional genetic algorithm.