Deformable models for texture segmentation

Segments a textured image by evolving a curve using a dictionary-based texture representation

Deformable models for segmenting textured images, including snakes, single-phase level-sets and multi-phase level-sets. Dictionary-based texture representation is the same for all models.
The methods are described in:
- "Dictionary snakes". Anders Bjorholm Dahl and Vedrana Andersen Dahl. 22nd International Conference on Pattern Recognition (ICPR), IEEE, 2014. http://www.computer.org/csdl/proceedings/icpr/2014/5209/00/5209a142-abs.html
- “Dictionary Based Image Segmentation". Anders Bjorhom Dahl and Vedrana Andersen Dahl. 19th Scandinavian Conference on Image Analysis (SCIA), Springer, 2015.
http://link.springer.com/chapter/10.1007%2F978-3-319-19665-7_3

After downloading the code, in texture_functions folder run compile_mex_functions once, and try demo scripts in the main folder. A set of 12 images is provided for testing.

Dictionary settings to experiment with: patch_size is a size of dictionary patches, relates to the texture type; branching_factor and number_layers relate to the size of the dictionary (larger dictionary provides better representation but slower computation). Curve evolution settings should be rather self-explanatory.