Active Contour Model Based on Local and Global Image Information

In this paper, a new hybrid active contour model for image segmentation was proposed. In our model, we define a new region-based signed pressure force (SPF) function which combines the local and the global image information. It can stop the contour at the weak or blurred edges. The proposed model can not only detect the boundaries automatically, but also is robust to noise and not sensitive to the initialization of the contour. In addition, the model can segment all objects in image domain. Compared with the other active contour model, the proposed model can be simply implemented by the finite difference scheme. Experiments on the synthetic and medical images demonstrate that the proposed active contour model can efficiently segment the objects well.