#!/usr/bin/env python
# vim: set fileencoding=utf-8 :
# Manuel Guenther <Manuel.Guenther@idiap.ch>
import bob.ip.gabor
import bob.io.base
import numpy
import math
from .Tool import Tool
[docs]class GaborJets (Tool):
"""Tool chain for computing Gabor jets, Gabor graphs, and Gabor graph comparisons"""
def __init__(
self,
# parameters for the tool
gabor_jet_similarity_type,
multiple_feature_scoring = 'max_jet',
# some similarity functions might need a GaborWaveletTransform class, so we have to provide the parameters here as well...
gabor_directions = 8,
gabor_scales = 5,
gabor_sigma = 2. * math.pi,
gabor_maximum_frequency = math.pi / 2.,
gabor_frequency_step = math.sqrt(.5),
gabor_power_of_k = 0,
gabor_dc_free = True
):
# call base class constructor
Tool.__init__(
self,
gabor_jet_similarity_type = gabor_jet_similarity_type,
multiple_feature_scoring = multiple_feature_scoring,
gabor_directions = gabor_directions,
gabor_scales = gabor_scales,
gabor_sigma = gabor_sigma,
gabor_maximum_frequency = gabor_maximum_frequency,
gabor_frequency_step = gabor_frequency_step,
gabor_power_of_k = gabor_power_of_k,
gabor_dc_free = gabor_dc_free,
multiple_model_scoring = None,
multiple_probe_scoring = None
)
# the Gabor wavelet transform; used by (some of) the Gabor jet similarities
gwt = bob.ip.gabor.Transform(
number_of_scales = gabor_scales,
number_of_directions = gabor_directions,
sigma = gabor_sigma,
k_max = gabor_maximum_frequency,
k_fac = gabor_frequency_step,
power_of_k = gabor_power_of_k,
dc_free = gabor_dc_free
)
# jet comparison function
self.m_similarity_function = bob.ip.gabor.Similarity(gabor_jet_similarity_type, gwt)
# how to proceed with multiple features per model
self.m_jet_scoring = {
'average_model' : None, # compute an average model
'average' : numpy.average, # compute the average similarity
'min_jet' : numpy.min, # for each jet location, compute the minimum similarity
'max_jet' : numpy.max, # for each jet location, compute the maximum similarity
'med_jet' : numpy.median, # for each jet location, compute the median similarity
'min_graph' : numpy.average, # for each model graph, compute the minimum average similarity
'max_graph' : numpy.average, # for each model graph, compute the maximum average similarity
'med_graph' : numpy.average, # for each model graph, compute the median average similarity
}[multiple_feature_scoring]
self.m_graph_scoring = {
'average_model' : None, # compute an average model
'average' : numpy.average, # compute the average similarity
'min_jet' : numpy.average, # for each jet location, compute the minimum similarity
'max_jet' : numpy.average, # for each jet location, compute the maximum similarity
'med_jet' : numpy.average, # for each jet location, compute the median similarity
'min_graph' : numpy.min, # for each model graph, compute the minimum average similarity
'max_graph' : numpy.max, # for each model graph, compute the maximum average similarity
'med_graph' : numpy.median, # for each model graph, compute the median average similarity
}[multiple_feature_scoring]
[docs] def enroll(self, enroll_features):
"""Enrolls the model by computing an average graph for each model"""
assert len(enroll_features)
if self.m_jet_scoring is not None:
return enroll_features
# compute average model
return [bob.ip.gabor.Jet([enroll_features[g][n] for g in range(len(enroll_features))], normalize=True) for n in range(len(enroll_features[0]))]
[docs] def save_model(self, model, model_file):
f = bob.io.base.HDF5File(model_file, 'w')
if self.m_jet_scoring is None:
# only one averaged model
bob.ip.gabor.save_jets(model, f)
else:
# several model graphs
f.set("NumberOfModels", len(model))
for g in range(len(model)):
name = "Model" + str(g+1)
f.create_group(name)
f.cd(name)
bob.ip.gabor.save_jets(model[g], f)
f.cd("..")
f.close()
[docs] def read_model(self, model_file):
f = bob.io.base.HDF5File(model_file)
if self.m_jet_scoring is None:
# only one graph
assert not f.has_key("NumberOfModels")
return bob.ip.gabor.load_jets(f)
else:
# several graphs
count = f.get("NumberOfModels")
model = []
for g in range(count):
name = "Model" + str(g+1)
f.cd(name)
model.append(bob.ip.gabor.load_jets(f))
f.cd("..")
return model
[docs] def read_probe(self, probe_file):
return bob.ip.gabor.load_jets(bob.io.base.HDF5File(probe_file))
[docs] def score(self, model, probe):
"""Computes the score of the probe and the model"""
if self.m_jet_scoring is None:
# compute sum of Gabor jet similarities between averaged model graph and probe graph
return numpy.average([self.m_similarity_function(model[n], probe[n]) for n in range(len(model))])
else:
# compute all Gabor jet similarities
scores = [[self.m_similarity_function(model[c][n], probe[n]) for n in range(len(model[0]))] for c in range(len(model))]
# for each jet location, compute the desired score averaging
return self.m_graph_scoring(self.m_jet_scoring(scores, axis=0))
[docs] def score_for_multiple_probes(self, model, probes):
"""This function computes the score between the given model graph(s) and several given probe graphs."""
if self.m_jet_scoring is None:
# compute sum of Gabor jet similarities between averaged model graph and probe graphs
return numpy.average([self.m_similarity_function(model[n], probes[p][n]) for n in range(len(model)) for p in range(len(probes))])
else:
# compute all Gabor jet similarities
scores = [[self.m_similarity_function(model[c][n], probes[p][n]) for n in range(len(model[0]))] for p in range(len(probes)) for c in range(len(model))]
# for each jet location, compute the desired score averaging
return self.m_graph_scoring(self.m_jet_scoring(scores, axis=0))
