- Remember to put 'test_images' to right path.....
Package for classic image processing techniques:
-
Thresholding package: containing Ostu and Reddi thresholding.
-
Classification package and Param_estimate package: using Maximum likelihood estimate, estimating the bias field (and the tissue classification of an MR image) based on maximum likelihood parameter estimation using the EM algorithm. (see Van Leemput's paper: Automated Model-Based Bias Field Correction of MR Images of the Brain)
-
Point_registration package: Containing point based registration using Optimization apporach (fminsearch) and ICP (see Dr. Fitzpatrick's paper)
-
Nonrigid_registration package: using thin-plate spline,
- Intensity based registration package: using Mutual information (joint-entropy):
-Snake shape package: see paper Snake: Active Contour Models. A segmentation technique using snake models, minimizing the energy function defined by internal and external force which is driven by the edge map (the target contour). Using three types of forces: gradient map, distance map and GVF (Gradient Vector Flow: A New External Force for Snakes),
- demon_matching package: using demons algorithm (Thirion's paper: Image matching as a diffusion process: an analogy with Maxwell’s demons) with image pyramid to perform the segmentation. An atlas-based segmentation of the ventricles, the intracranial cavity and the head contours of human brains:
- Active shape models package: containing hand segmentation using PCA algorithm.
4 of the main eigen vecs of the mean hand space
red hand is the target hand contour and the green contours are the initial, first iteration and the final fitting hand results
