Firstly

tracking model of moving objects in images is established based on Bayesian theory. Secondly

image regional information is described by Gaussian distribution. At the same time

the tracking model is matched to region-based active contours through analyzing

and then this problem is converted into the minimizing of energy. Lastly

level set is used to solve the PDE equation

which can process topology naturally. The result shows that the proposed method not only could track multi-objects and approach to the objects contours

but also does well when the moving objects deforms topologically.