Aiming at problem of low speed of convergence

early mature of standard genetic algorithm（ GA） occur in optimizing Otsu method to search for threshold of image

an improved GA for image segmentation is proposed.According to different evolution generation and individual fitness

this improved GA can adjust strategies of elite selection and genetic operator dynamically

so it not only can speed up the convergence speed

but also can get image segmentation threshold with stable range and keep diversity of population. This algorithm is used in medical image segmentation

and experimental results show that the algorithm can be used for segmentation of medical image and effect is obvious.