def _pairwise_similarity_matrix(input_polydata, threshold, sigma,
number_of_jobs=3, landmarks=None):
""" Internal convenience function available to clustering
routines.
Computes similarity matrix (nxn) for all n fibers in input
polydata. Calls function _pairwise_distance_matrix first.
For distance computation and conversion to similarity uses
parameters threshold and sigma.
"""
distances = _pairwise_distance_matrix(input_polydata, threshold,
number_of_jobs, landmarks)
# similarity matrix
sigmasq = sigma * sigma
similarity_matrix = similarity.distance_to_similarity(distances, sigmasq)
# sanity check that on-diagonal elements are all 1
#print "This should be 1.0: ", numpy.min(numpy.diag(similarity_matrix))
#print numpy.min(numpy.diag(similarity_matrix)) == 1.0
# test
if __debug__:
# this tests that on-diagonal elements are all 1
test = numpy.min(numpy.diag(similarity_matrix)) == 1.0
if not test:
print "<cluster.py> ERROR: On-diagonal elements are not all 1.0."
raise AssertionError
return similarity_matrix
def _rectangular_similarity_matrix(input_polydata_n, input_polydata_m, threshold, sigma,
number_of_jobs=3, landmarks_n=None, landmarks_m=None, distance_method='Hausdorff',
bilateral=False):
""" Internal convenience function available to clustering
routines.
Computes similarity matrix (nxn) for all n fibers in input
polydata. Calls function _pairwise_distance_matrix first.
For distance computation and conversion to similarity uses
parameters threshold and sigma.
"""
distances = _rectangular_distance_matrix(input_polydata_n, input_polydata_m, threshold,
number_of_jobs, landmarks_n, landmarks_m, distance_method, bilateral=bilateral)
if distance_method == 'StrictSimilarity':
similarity_matrix = distances
else:
# similarity matrix
sigmasq = sigma * sigma
similarity_matrix = similarity.distance_to_similarity(distances, sigmasq)
return similarity_matrix
