An Effective Level Set Image Segmentation by Joint Local Kernelized Model and Global Chan-Vese Model

This study presents a novel level set method for image segmentation by means of local kernel mapping and piecewise constant modeling of the image data to deal with image segmentation with intensity non-homogeneity problem. The proposed method adopts local kernel mapping to enhance the discriminative ability to delineate nonlinear decision boundaries between classes. In addition, our approach method embeds a Chan-Vese model into the energy function, which not only can enhance the robustness against noise but also make our approach less sensitive to the localization of the initial contour. We verified the results of the method by a comparative study over a large number of experiments on synthetic and real images. The experiments demonstrate that our method is efficient and robust for segmenting images with intensity inhomogeneity, noise images and texture images.