Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations
Journal of Computational Physics
Scale-Space and Edge Detection Using Anisotropic Diffusion
IEEE Transactions on Pattern Analysis and Machine Intelligence
Region-based strategies for active contour models
International Journal of Computer Vision
A Multiphase Level Set Framework for Image Segmentation Using the Mumford and Shah Model
International Journal of Computer Vision
Brain MR Image Segmentation Using Local and Global Intensity Fitting Active Contours/Surfaces
MICCAI '08 Proceedings of the 11th international conference on Medical Image Computing and Computer-Assisted Intervention - Part I
Snakes, shapes, and gradient vector flow
IEEE Transactions on Image Processing
IEEE Transactions on Image Processing
Minimization of Region-Scalable Fitting Energy for Image Segmentation
IEEE Transactions on Image Processing
Hi-index | 0.00 |
This paper proposes a new Globally Optimal and Region Scalable Chan-Vese model (GRCV model), which combines the advantages of both local and global intensity information. Intensity information in local regions is emphasized by adding a kernel function in the data fitting term, which thereby enables the proposed model to extract the fine structure in local region. The global information is also considered by adding one data fitting term in Piecewise Constant (PC) model to guarantee that proposed model can successfully segment complete target regions from intensity inhomogeneous images. We give the selection rules of combining the different terms under different target and background intensities: when the target is intensity inhomogeneous while the background is homogeneous, we choose the data fitting term which effects on the outside region of the evolution contour; conversely, we choose the other term. Experiments have been done on different images to compare the effectiveness of our methods with that of the classical PC model, Li's Local Binary Fitting (LBF) model, and Wang's Local and Global Intensity Fitting (LGIF) model. Comparison results show that our model obtains more satisfactory segmentation results. Moreover, it is robust to the curve initialization and noise in images.