Snakes, or active contours, are used extensively in medical image or geographic image processing and analysis. However, initialization problem and concave boundary problem seriously affect the performance of snake model. This paper presents a snake model based on Maximum Gravity Algorithm (MGA), a new global optimization algorithm which has the merits of less individuals, simple calculation, robustness and high performance. In our method, we choose energy formula of traditional snake model as the objective function, MGA as the optimization algorithm. The experiment results show that our algorithm can overcome the shortcomings of traditional snake model, and image segmentation accuracy is high.