Improved segmentation model combining region and edge information for inhomogeneous images

In this paper we propose an improved image segmentation model combining the region and edge information for inhomogeneous images. First, we define a new energy functional in a variational level set formulation based on the region information, including the local and global intensity fitting terms. Then we incorporate the edge information into the energy functional by adding a non-negative edge detector function to detect boundaries more easily. We apply a weight function to control the influence of the local and global intensity information dynamically. Therefore, the proposed model can segment more general images more accurately, including images with intensity inhomogeneity. Finally, the special structure of the newly defined energy functional ensures that we can apply the split Bregman method to minimize it much more efficiently. We have applied our model to synthetic and real images and numerical results have demonstrated the high efficiency of the improved model.