A Class Of Image Segmentation Models Based On Continuous Maximum Flow Methods

In this paper,the improved model of continuous max-flow image segmentation and the numerical realization of the fast algorithm of the model are mainly studied.In image processing,image segmentation through the CV model is time-consuming and needs many iterations.When we are studying the CV model,we use the idea of duality to carry out a series of derivation of the original energy functional of the CV model to simplify and obtain the energy functional of the continuous max-flow model,which is explained in graph theory.The segmentation result of continuous maximum flow model is susceptible to the influence of parameters and step size.Over segmentation can lead to a large number of step-effect artifacts and the texture features are not obvious.Firstly,the images are processed by OTSU method,then the continuous maximum flow is used for secondary segmentation.On this basis,we have improved the parameters of the continuous max-flow model.The experimental results show that our new algorithm has more advantages in speed and segmentation effect.