Evolutionary Morphing and Shape Distance - PowerPoint PPT Presentation

About This Presentation
Title:

Evolutionary Morphing and Shape Distance

Description:

Evolutionary Morphing and Shape Distance Nina Amenta Computer Science, UC Davis – PowerPoint PPT presentation

Number of Views:138
Avg rating:3.0/5.0
Slides: 37
Provided by: Nina266
Category:

less

Transcript and Presenter's Notes

Title: Evolutionary Morphing and Shape Distance


1
Evolutionary Morphing and Shape Distance
  • Nina Amenta
  • Computer Science, UC Davis

2
Collaborators
  • Physical Anthropology
  • Eric Delson, Steve Frost, Lissa Tallman, Will
    Harcourt-Smith
  • Morphometrics
  • F. James Rohlf
  • Computer Science and Math
  • Katherine St. John, David Wiley, Deboshmita
    Ghosh, Misha Kazhdan, Owen Carmichael, Joel Hass,
    David Coeurjolly

3
Outline
  • Application of 3D Procrustes tangent space
    analysis in primate evolution
  • Some issues with the shape space
  • An idea

4
Evolutionary Trees
5
Computing Trees
Trees on extant species come from genomic data.
Papio Macaca Cercocebus Cercopithecus Allenopithec
us
Tree inference method
6
Estimating morphology
  • Using 3D data for extant species, and tree,
    estimate cranial shapes for the hypothetical
    ancestors.

3D input data
7
Estimating morphology
  • Generalized least-squares, covariance matrix
    derived from weighted tree edges.

8
Evolutionary Morphing
9
Fossils
  • Genomic trees dont include fossils.
  • Primates 200 extinct genera, 60 extant.
  • Fossils have to be added based on shape and
    meta-data.

10
Fossil Restoration
symmetrization
reflection
fossil
11
Sahelanthropos
12
Fossil Restoration
TPS
restored fossil
template surface
reconstructed specimen
13
Improve Estimated Morphology
synthetic basal node
repaired Victoriapithecus
14
Improve Estimated Morphology
improved basal node
repaired Victoriapithecus
15
Parapapio, a more recent fossil
Template is root of subtree where we believe it
falls
16
Placement of Parapaio
17
User-defined landmarks
We optimize for correspondence only within
surface patches (Bookstein sliding, does not
work
well).
Our users want to specify or edit landmarks, but
more automation is clearly needed.
18
Procrustes Distance
  • DEuc(A,B) Euclidean distance in R3n
  • Choose transformation T (scale, trans, rot)
    producing minimum DEuc
  • DProc(A,B) min DEuc(T(A), B)
  • T
  • We work in Euclidean tangent space.

19
Example
20
Features are not aligned
..even starting with optimal correspondence.
Procrustes distance emphasizes big change, misses
similarity of parts.
21
Features are not aligned
Changing the details might even reduce DProc.
22
Features are not aligned
Optimizing correspondence under DProc will not
lead to intuitively better correspondence.
23
Complex Shapes
All parts cannot be simultaneously aligned by
linear deformations. Deformation really is
non-linear.
24
Edge-length Distance
Proposal represent correspondence as
corresponding triangle meshes instead of
corresponding point samples.
25
Edge-length Distance
  • Li is Euclidean length of edge ei
  • Shape feature vector v is (L1 Lk)
  • DEL DEuc(v(A), v(B))
  • This represents a mesh as a discrete metric set
    of lengths on a triangulated graph, respecting
    the triangle inequality

26
Information Loss
  • In 2D, this does not make much sense.
  • But in 3D, almost all triangulated polyhedra are
    rigid. So a discrete metric has a finite number
    of rigid realizations.

27
Not a New Idea
  • Euclidean Distance Matrix Analysis, Lele and
    Richtmeier, 2001 use the complete distance
    matrix as shape rep.
  • Truss metrics include only enough edges to
    get rigidity.

28
Quote
  • the arbitrary choice of a subset of linear
    distances could accentuate the influence of
    certain linear distances in the comparison of
    forms, while masking the influence of others. -
    Richtsmeier, Deleon, and Lele, 2002.
  • Not an issue in R3!

29
Nice Properties
  • Rotation and translation invariant
  • Invariant to rotations and translations of parts
    (isometries).
  • Any convex combination of specimens gives another
    vector of Li obeying triangle inequalities. So
    we can do statistics in a convex region of
    Euclidean space.

30
Scale
  • Can normalize to produce scale invariance, as
    with Procrustes distance.
  • Choosing scale so that S Li 1 keeps all
    specimens in a linear subspace.

31
Degrees of Freedom
  • Dimension of Kendall shape space is 3n-7
  • Number of edges for a triangulated object
    homeomorphic to a sphere is 3n-6
    (Eulertriangulation constraints), -1 for scale
    3n-7

32
Scale
  • But this does not solve the problem of matching
    parts getting different scales.
  • What if we apply local scale factors at each
    vertex?

33
Local Scale?
  • We could add a scale factor at each vertex,
    producing a discrete conformal representation
    (Springborn, Schoeder, Bobenko, Pinkall)but this
    has way too many degrees of freedom.
  • Q1 How to incorporate the right amount of local
    scale?

34
Drawback
  • Isometric surfaces have distance zero.
  • Complicates reconstruction of interpolated
    shapes. Q2.

35
More Questions
  • Q3 Given a discrete metric formed as a convex
    combination of specimens, how to choose the right
    3D realization for visualization?
  • Q4 How to optimize correspondence so as to
    minimize DEL? How to weight by area?

36
Thank you!
37
Correspondence-base Metrics
  • Correspondence diffeomorphism or sampling
    therof. Usually point samples.

38
Invariance
  • to some set of transformations of the input (eg,
    rot, trans, scale).
  • DRMS (sum-squared difference) is notbut it gives
    a Euclidean shape space.
  • DTPS (bending energy) isbut does not give a
    Euclidean shape space, and has other problems.

39
Commercialization
40
Shape Metric
  • Requires a shape metric.
  • Much easier in Euclidean shape-space.

41
Shape Is Invariant
  • ...to transformations of the input (rot, trans,
    scale, correspondence).

n landmarks
B
A
42
Tangent Space
  • DProc does not give a Euclidean shape space.
  • But computing the Karcher mean M and aligning all
    specimens to M does gives a nearby Euclidean
    shape space.

43
Invariance is not enough
  • We could remove nuisance parameters from lots of
    metrics, achieving invariance, and we would still
    not necessarily be measuring shape.
  • Consistency is important, but does not
    necessarily imply we are measuring shape
    distance.

44
Philosophy
  • But, we feel like DProc should be a shape metric
    because
  • DEuc is a good metric for shapes distorted by
    noise
  • The transformation T that we find seems correct
    (except for correspondence)
  • It works OK in applications

45
Where did we go wrong?
  • In the initial choice of DE

What should we do?
Take the derivative!
Write a Comment
User Comments (0)
About PowerShow.com