Abstract: In recent years,Geometric Active Contour model has widely been used in image segmentation.CV model can divide the target and the background into two piecewise smooth areas,drive the contour curve moving to the target edge by minimizing the Mumford-Shah function.CV model is not sensitive to the initial curve location and has better noise immunity,but easy make the mistake such as the wrong target recognition or loss some edges when it used for low contrast image segmentation.In 2008,Li,C.M. et.al presented the Region Salable Fitting model by using the Gaussian function as the local region kernel function to fit the energy functional.RSF model can segment the low contrast image,but its contour curve may converge to the pseudo edge and sensitive to the noise and initial position.In this paper,we join the phase contingency function as the edge stop item into RSF model,convert the solution into the global convex segmentation energy functional.We call this model as PRSF.Some experimental results show that PRSF model can integrate the features of area and edge active contour effectively,it is not sensitive to the noise and the initial position,and can extract the edge of weak target in the low contrast.We use PRSF model to segment the meteorological target with Split Bergman fast computing,the algorithm of time execution efficiency is higher.PRSF model can be used for real-time target tracking calculation.