A MATLAB implementation of the Continuous Procrustes Distance
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
demos
samples
utils
viewer
.gitignore
README.md
cPDemoPNAS.m
sandbox.m

README.md

DEMO SCRIPTS

  1. "cPDemoPNAS.m" demonstrates the Continuous Procrustes Analysis algorithm on the PNAS lemur teeth data set.
  2. "cPMSTDemoTeeth.m" demonstrates some spanning-tree-based improvements of the Continuous Procrustes Analysis algorithm on the PNAS lemur teeth data set.

WORKFLOW DESCRIPTION FOR COLLECTION ANALYSIS

CONTINUOUS PROCRUSTES MINIMAL SPANNING TREE (CP-MST)

  1. Run "cluster_flatten.m" to generate samples (saved under results/).

  2. Run "cluster_cPdist.m" to generate pairwise comparision results (saved under rslts/; cluster scripts and debugging information stored under cluster/).

  3. Run "cPProcessRslts.m" to separate results from rslts/. The results are stored in results/; typically cPdistMatrix.mat, cPmapsMartix.mat, lmkMSEMatrix.m, TextureCoords1/, TextureCoords2/ are generated and saved.

  4. Move rslts/, TextureCoords1/, TextureCoords2/ to other locations (taking too much space).

  5. Run "cluster_Imprdist.m" to try different types of improvements.

  6. Run "ImprProcessRslts.m" to separate results from rslts/. The results are stored in results/; typically cP[ImprType]distMatrix.mat, cP[ImprType]MapsMartix.mat, cP[ImprType]lmkMSEMatrix.m, TextureCoords1/, TextureCoords2/ are generated and saved.

  7. As in Step 4, move rslts/, TextureCoords1/, TextureCoords2/ to other locations for future reference.


Tingran Gao, Duke University Feb 3, 2015