Home

Level Set Methods / Active Contour Models for Image Segmentation

Edge based models:
  1. Level Set Evolution without Re-initialization: A New Variational Formulation, CVPR'05 (oral presentation). Click here to see some demos and download Matlab code for this method. C code for this method is available by email request. This method was improved in my IEEE TIP paper ( try the code). The new paper and the source code will be available here. Note: In the new method, the level set function is intrinsically maintained as an approximate signed distance only near the zero level set.
  2. Fast narrow band implementation of the above method, (pdf, demo code).
  3. Segmentation of external force field for a snake model, CVPR'05.

 

Region based models:
  • Implicit Active Contours Driven by Local Binary Fitting Energy, CVPR'07. (Matlab code: LBF_v0)
  • Minimization of Region-Scalable Fitting Energy for Image Segmentation, to appear in IEEE Trans. Image Processing.

Implementations and further developments:

  1. LBF_v0.1 (.rar or .zip) for Matlab 7.0 or higher versions in Windows.
  2. LBF_v0.5 (with an auxiliary global term) (MICCAI 2008)
  3. LBF_1.0 (simultaneous segmentation and bias field estimation) (MICCAI 2008, oral presentation)
  • Segmentation of brain MR images with intensity inhomogeneity. This model is able to simultaneously segment gray matter, white matter, cerebrospinal fluid (CSF), and the background. One page abstract to appear in ISMRM'08 (oral presentation).
3D versions:
Implementations of other level set/active contour methods: 

Note: More and updated code will be released in my main homepage.