def __init__(self, shapes, pca_variance_captured=0.9, gpa_tol=1e-7, gpa_max_iters=10000):
"""
Constructs the Active Shape Model based on the given list of Shapes.
:param shapes: A list of Shapes
:param pca_variance_captured: The fraction of variance to be captured by the shape model
:param gpa_tol: tol: The convergence threshold for gpa
(Default: 1e-7)
:param gpa_max_iters: The maximum number of iterations
permitted for gpa (Default: 10000)
"""
self._aligned_shapes = AlignedShapeList(shapes, gpa_tol, gpa_max_iters)
self._model = ModedPCAModel(self._aligned_shapes.raw(), pca_variance_captured)
class ShapeModel:
""" An Active Shape Model based on
Cootes, Tim, E. R. Baldock, and J. Graham.
"An introduction to active shape models."
Image processing and analysis (2000): 223-248.
Attributes:
_aligned_shapes An AlignedShapeList containing
the training shapes
_model The underlying modedPCA Model
Authors: David Torrejon and Bharath Venkatesh
"""
def __init__(self, shapes, pca_variance_captured=0.9, gpa_tol=1e-7, gpa_max_iters=10000):
"""
Constructs the Active Shape Model based on the given list of Shapes.
:param shapes: A list of Shapes
:param pca_variance_captured: The fraction of variance to be captured by the shape model
:param gpa_tol: tol: The convergence threshold for gpa
(Default: 1e-7)
:param gpa_max_iters: The maximum number of iterations
permitted for gpa (Default: 10000)
"""
self._aligned_shapes = AlignedShapeList(shapes, gpa_tol, gpa_max_iters)
self._model = ModedPCAModel(self._aligned_shapes.raw(), pca_variance_captured)
def aligned_shapes(self):
"""
Returns the gpa aligned shapes
:return: A list of Shape objects containing the aligned shapes
"""
return self._aligned_shapes.shapes()
def mean_shape(self):
"""
Returns the mean shape of the model
:return: A Shape object containing the mean shape
"""
return self._aligned_shapes.mean_shape()
def modes(self):
"""
Returns the number of modes of the model
:return: the number of modes
"""
return self._model.modes()
def generate_shape(self, factors):
"""
Generates a shape based on a vector of factors of size
equal to the number of modes of the model, with element
values between -1 and 1
:param factors: A vector of size modes() with values
between -1 and 1
:return: A Shape object containing the generated shape
"""
return Shape(
self.mean_shape().raw() + Shape.from_collapsed_shape(self._model.generate_deviation(factors)).raw())
def mode_shapes(self, m):
"""
Returns the modal shapes of the model (Variance limits)
:param m: A list of Shape objects containing the modal shapes
"""
if m < 0 or m >= self.modes():
raise ValueError('Number of modes must be within [0,modes()-1]')
factors = np.zeros(self.modes())
mode_shapes = []
for i in range(-1, 2):
factors[m] = i
mode_shapes.append(self.generate_shape(factors))
return mode_shapes