| fgProcrustes {Momocs} | R Documentation |
Full Generalized Procrustes alignment between shapes
Description
Directly borrowed from Claude (2008), called there the fgpa2 function.
Usage
fgProcrustes(x, tol, coo)
Arguments
x |
an array, a list of configurations, or an Out, Opn or Ldk object |
tol |
numeric when to stop iterations |
coo |
logical, when working on |
Details
If performed on an Out or an Opn object, will try to use the $ldk slot,
if landmarks have been previousy defined, then (with a message) on the $coo slot,
but in that case, all shapes must have the same number of coordinates (coo_sample may help).
Value
a list with components:
-
rotatedarray of superimposed configurations -
iterationnumbernumber of iterations -
Qconvergence criterion -
Qifull list of Q -
Qddifference between successive Q -
interproc.distminimal sum of squared norms of pairwise differences between all shapes in the superimposed sample -
mshapemean shape configuration -
cent.sizevector of centroid sizes.
or an Out, Opn or an Ldk object.
Note
Slightly less optimized than procGPA in the shapes package (~20% on my machine).
Will be optimized when performance will be the last thing to improve!
Silent message and progress bars (if any) with options("verbose"=FALSE).
References
Claude, J. (2008). Morphometrics with R. Analysis (p. 316). Springer.
See Also
Other procrustes functions:
fProcrustes(),
fgsProcrustes(),
pProcrustes()
Examples
# on Ldk
w <- wings %>% slice(1:5) # for the sake of speed
stack(w)
fgProcrustes(w, tol=0.1) %>% stack()
# on Out
h <- hearts %>% slice(1:5) # for the sake of speed
stack(h)
fgProcrustes(h) %>% stack()