![Page 1: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/1.jpg)
Invariant Histograms andSignatures for Object
Recognition and SymmetryDetection
Peter J. Olver
University of Minnesota
http://www.math.umn.edu/! olver
North Carolina, October, 2011
![Page 2: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/2.jpg)
References
Boutin, M., & Kemper, G., On reconstructing n-pointconfigurations from the distribution of distances orareas, Adv. Appl. Math. 32 (2004) 709–735
Brinkman, D., & Olver, P.J., Invariant histograms,Amer. Math. Monthly 118 (2011) 2–24.
Calabi, E., Olver, P.J., Shakiban, C., Tannenbaum, A., &Haker, S., Di!erential and numerically invariantsignature curves applied to object recognition, Int.J. Computer Vision 26 (1998) 107–135
![Page 3: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/3.jpg)
The Distance Histogram
Definition. The distance histogram of a finite set ofpoints P = {z1, . . . , zn} " V is the function
!P (r) = #!
(i, j)""" 1 # i < j # n, d(zi, zj) = r
#.
![Page 4: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/4.jpg)
The Distance Set
The support of the histogram function,
supp !P = "P " R+
is the distance set of P .
Erdos’ distinct distances conjecture (1946):
If P " Rm, then # "P $ cm,! (# P )2/m!!
![Page 5: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/5.jpg)
Characterization of Point Sets
Note: If $P = g · P is obtained from P " Rm by arigid motion g % E(n), then they have the samedistance histogram: !P = !$P .
Question: Can one uniquely characterize, up to rigidmotion, a set of points P{z1, . . . , zn} " Rm by itsdistance histogram?
=& Tinkertoy problem.
![Page 6: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/6.jpg)
Yes:
! = 1, 1, 1, 1,'
2,'
2.
![Page 7: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/7.jpg)
No:
Kite Trapezoid
! ='
2,'
2, 2,'
10,'
10, 4.
![Page 8: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/8.jpg)
No:
P = {0, 1, 4, 10, 12, 17}
Q = {0, 1, 8, 11, 13, 17}" R
! = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 16, 17
=& G. Bloom, J. Comb. Theory, Ser. A 22 (1977) 378–379
![Page 9: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/9.jpg)
Theorem. (Boutin–Kemper) Suppose n # 3 or n $ m + 2.Then there is a Zariski dense open subset in the space of npoint configurations in Rm that are uniquely characterized,up to rigid motion, by their distance histograms.
=& M. Boutin, G. Kemper, Adv. Appl. Math. 32 (2004) 709–735
![Page 10: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/10.jpg)
Limiting Curve Histogram
![Page 11: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/11.jpg)
Limiting Curve Histogram
![Page 12: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/12.jpg)
Limiting Curve Histogram
![Page 13: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/13.jpg)
Sample Point HistogramsCumulative distance histogram: n = #P :
#P (r) =1
n+
2
n2
%
s"r
!P (s) =1
n2#
!(i, j)
""" d(zi, zj) # r#
,
Note!(r) = 1
2 n2[ #P (r) ( #P (r ( ") ] " ) 1.
Local distance histogram:
#P (r, z) =1
n#
!j
""" d(z, zj) # r#
=1
n#(P * Br(z))
Ball of radius r centered at z:
Br(z) = { v % V | d(v, z) # r }
Note:
#P (r) =1
n
%
z #P
#P (r, z) =1
n2
%
z #P
#(P * Br(z)).
![Page 14: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/14.jpg)
Limiting Curve Histogram Functions
Length of a curve
l(C) =&
Cds < +
Local curve distance histogram function z % V
hC(r, z) =l(C * Br(z))
l(C)
=& The fraction of the curve contained in the ball of radius rcentered at z.
Global curve distance histogram function:
HC(r) =1
l(C)
&
ChC(r, z(s)) ds.
![Page 15: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/15.jpg)
Convergence
Theorem. Let C be a regular plane curve. Then, for bothuniformly spaced and randomly chosen sample pointsP " C, the cumulative local and global histograms convergeto their continuous counterparts:
#P (r, z) (, hC(r, z), #P (r) (, HC(r),
as the number of sample points goes to infinity.
![Page 16: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/16.jpg)
Square Curve Histogram with Bounds
![Page 17: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/17.jpg)
Kite and Trapezoid Curve Histograms
![Page 18: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/18.jpg)
Histogram–Based Shape Recognition500 sample points
Shape (a) (b) (c) (d) (e) (f)
(a) triangle 2.3 20.4 66.9 81.0 28.5 76.8
(b) square 28.2 .5 81.2 73.6 34.8 72.1
(c) circle 66.9 79.6 .5 137.0 89.2 138.0
(d) 2 - 3 rectangle 85.8 75.9 141.0 2.2 53.4 9.9
(e) 1 - 3 rectangle 31.8 36.7 83.7 55.7 4.0 46.5
(f) star 81.0 74.3 139.0 9.3 60.5 .9
![Page 19: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/19.jpg)
Curve Histogram Conjecture
Two su$ciently regular plane curves C
and $C have identical global distance
histogram functions, so HC(r) = H$C(r)
for all r $ 0, if and only if they are
rigidly equivalent: C . $C.
![Page 20: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/20.jpg)
“Proof Strategies”
• Show that any polygon obtained from (densely) discretizing acurve does not lie in the Boutin–Kemper exceptional set.
• Polygons with obtuse angles: taking r small, one can recover(i) the set of angles and (ii) the shortest side length fromHC(r). Further increasing r leads to further geometricinformation about the polygon . . .
• Expand HC(r) in a Taylor series at r = 0 and show that thecorresponding integral invariants characterize the curve.
![Page 21: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/21.jpg)
Taylor Expansions
Local distance histogram function:
LhC(r, z) = 2r + 112 $
2 r3 +'
140 $$ss + 1
45 $2s + 3
320 $4
(r5 + · · · .
Global distance histogram function:
HC(r) =2r
L+
r3
12L2
)
C$2 ds +
r5
40L2
)
C
'38 $
4 ( 19 $
2s
(ds + · · · .
![Page 22: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/22.jpg)
Space Curves
Saddle curve:
z(t) = (cos t, sin t, cos 2 t), 0 # t # 2%.
Convergence of global curve distance histogram function:
![Page 23: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/23.jpg)
SurfacesLocal and global surface distance histogram functions:
hS(r, z) =area (S * Br(z))
area (S), HS(r) =
1
area (S)
& &
ShS(r, z) dS.
Convergence for sphere:
![Page 24: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/24.jpg)
Area Histograms
Rewrite global curve distance histogram function:
HC(r) =1
L
)
ChC(r, z(s)) ds =
1
L2
)
C
)
C&r(d(z(s), z(s$)) ds ds$
where &r(t) =
*1, t # r,
0, t > r,
Global curve area histogram function
AC(r) =1
L3
)
C
)
C
)
C&r(area (z(+s), z(+s$), z(+s$$)) d+s d+s$ d+s$$,
d+s — equi-a$ne arc length element L =&
Cd+s
Discrete cumulative area histogram
AP (r) =1
n(n ( 1)(n ( 2)
%
z %=z! %=z!!#P
&r(area (z, z$, z$$)),
Boutin & Kemper: the area histogram uniquely determinesgeneric point sets P " R2 up to equi-a$ne motion
![Page 25: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/25.jpg)
Area Histogram for Circle
' ' Joint invariant histograms — convergence???
![Page 26: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/26.jpg)
Triangle Distance Histograms
Z = (. . . zi . . .) " M — sample points on a subset M " Rn
(curve, surface, etc.)
Ti,j,k — triangle with vertices zi, zj, zk.
Side lengths:
((Ti,j,k) = ( d(zi, zj), d(zi, zk), d(zj, zk) )
Discrete triangle histogram:
S = ((T ) " K
Triangle inequality cone
K = { (x, y, z) | x, y, z $ 0, x + y $ z, x + z $ y, y + z $ x } " R3.
![Page 27: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/27.jpg)
Triangle Histogram Distributions
Circle Triangle Square
=& Madeleine Kotzagiannidis
![Page 28: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/28.jpg)
Practical Object Recognition
• Scale-invariant feature transform (SIFT) (Lowe)
• Shape contexts (Belongie–Malik–Puzicha)
• Integral invariants (Krim, Kogan, Yezzi, Pottman, . . . )
• Shape distributions (Osada–Funkhouser–Chazelle–Dobkin)Surfaces: distances, angles, areas, volumes, etc.
• Gromov–Hausdor! and Gromov-Wasserstein distances (Memoli)=& lower bounds
![Page 29: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/29.jpg)
Signature Curves
Definition. The signature curve S " R2 of a curve C " R2 is
parametrized by the two lowest order di!erential invariants
S =
* ,
$ ,d$
ds
- .
" R2
=& One can recover the signature curve from the Taylor
expansion of the local distance histogram function.
![Page 30: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/30.jpg)
Other Signatures
Euclidean space curves: C " R3
S = { ($ , $s , ) ) } " R3
• $ — curvature, ) — torsion
Euclidean surfaces: S " R3 (generic)
S =! '
H , K , H,1 , H,2 , K,1 , K,2
( #" R
3
• H — mean curvature, K — Gauss curvature
Equi–a!ne surfaces: S " R3 (generic)
S =! '
P , P,1 , P,2, P,11
( #" R
3
• P — Pick invariant
![Page 31: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/31.jpg)
Equivalence and Signature Curves
Theorem. Two regular curves C and C are equivalent:
C = g · C
if and only if their signature curves are identical:
S = S
=& object recognition
![Page 32: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/32.jpg)
Symmetry and Signature
Theorem. The dimension of the symmetry group
GN = { g | g · N " N }
of a nonsingular submanifold N " M equals the
codimension of its signature:
dimGN = dim N ( dimS
![Page 33: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/33.jpg)
Discrete Symmetries
Definition. The index of a submanifold N equalsthe number of points in N which map to a genericpoint of its signature:
*N = min!
# %!1{w}""" w % S
#
=& Self–intersections
Theorem. The cardinality of the symmetry group ofa submanifold N equals its index *N .
=& Approximate symmetries
![Page 34: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/34.jpg)
The Index
%
(,
N S
![Page 35: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/35.jpg)
=& Steve Haker
![Page 36: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/36.jpg)
![Page 37: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/37.jpg)
![Page 38: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/38.jpg)
Signature Metrics
• Hausdor!
• Monge–Kantorovich transport
• Electrostatic repulsion
• Latent semantic analysis (Shakiban)
• Histograms (Kemper–Boutin)
• Di!usion metric
• Gromov–Hausdor!
![Page 39: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/39.jpg)
Signatures
s
$
Classical Signature(,
Original curve$
$s
Di!erential invariant signature
![Page 40: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/40.jpg)
Signatures
s
$
Classical Signature(,
Original curve$
$s
Di!erential invariant signature
![Page 41: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/41.jpg)
Occlusions
s
$
Classical Signature(,
Original curve$
$s
Di!erential invariant signature
![Page 42: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/42.jpg)
The Ba"er Jigsaw Puzzle
![Page 43: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/43.jpg)
The Ba"er Solved
=& Dan Ho!
![Page 44: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/44.jpg)
Advantages of the Signature Curve
• Purely local — no ambiguities
• Symmetries and approximate symmetries
• Extends to surfaces and higher dimensionalsubmanifolds
• Occlusions and reconstruction
Main disadvantage: Noise sensitivity due todependence on high order derivatives.
![Page 45: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/45.jpg)
Noise Reduction
' Use lower order invariants to construct a signature:
• joint invariants
• joint di!erential invariants
• integral invariants
• topological invariants
• . . .
![Page 46: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/46.jpg)
Joint Invariants
A joint invariant is an invariant of the k-foldCartesian product action of G on M - · · ·- M :
I(g · z1, . . . , g · zk) = I(z1, . . . , zk)
A joint di!erential invariant or semi-di!erentialinvariant is an invariant depending on the derivativesat several points z1, . . . , zk % N on the submanifold:
I(g · z(n)1 , . . . , g · z(n)
k ) = I(z(n)1 , . . . , z
(n)k )
![Page 47: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/47.jpg)
Joint Euclidean Invariants
Theorem. Every joint Euclidean invariant is afunction of the interpoint distances
d(zi, zj) = / zi ( zj /
zi
zj
![Page 48: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/48.jpg)
Joint Equi–A!ne Invariants
Theorem. Every planar joint equi–a$ne invariant isa function of the triangular areas
[ i j k ] = 12 (zi ( zj) 0 (zi ( zk)
zi
zj
zk
![Page 49: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/49.jpg)
Joint Projective Invariants
Theorem. Every joint projective invariant is afunction of the planar cross-ratios
[ zi, zj, zk, zl, zm ] =A B
C D
A B
C
D
![Page 50: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/50.jpg)
• Three–point projective joint di!erential invariant— tangent triangle ratio:
[ 0 2!
0 ] [ 0 1!
1 ] [ 1 2!
2 ]
[ 0 1!
0 ] [ 1 2!
1 ] [ 0 2!
2 ]
z0 z1
z2
z0 z1
z2
![Page 51: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/51.jpg)
Joint Invariant Signatures
If the invariants depend on k points on a p-dimensional
submanifold, then you need at least
+ > k p
distinct invariants I1, . . . , I" in order to construct a syzygy.
Typically, the number of joint invariants is
+ = k m ( r = (#points) (dimM) ( dim G
Therefore, a purely joint invariant signature requires at least
k $r
m ( p+ 1
points on our p-dimensional submanifold N " M .
![Page 52: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/52.jpg)
Joint Euclidean Signature
z0 z1
z2z3
a
b
c d
e
f
![Page 53: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/53.jpg)
Joint signature map:
% : C&4 (, S " R6
a = / z0 ( z1 / b = / z0 ( z2 / c = / z0 ( z3 /
d = / z1 ( z2 / e = / z1 ( z3 / f = / z2 ( z3 /
=& six functions of four variables
Syzygies: &1(a, b, c, d, e, f) = 0 &2(a, b, c, d, e, f) = 0
Universal Cayley–Menger syzygy 1& C " R2
det
"""""""
2a2 a2 + b2 ( d2 a2 + c2 ( e2
a2 + b2 ( d2 2b2 b2 + c2 ( f2
a2 + c2 ( e2 b2 + c2 ( f2 2c2
"""""""= 0
![Page 54: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/54.jpg)
Joint Equi–A!ne Signature
Requires 7 triangular areas:
[ 0 1 2 ] , [ 0 1 3 ] , [ 0 1 4 ] , [ 0 1 5 ] , [ 0 2 3 ] , [ 0 2 4 ] , [ 0 2 5 ]
z0
z1
z2
z3
z4
z5
![Page 55: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/55.jpg)
Joint Invariant Signatures
• The joint invariant signature subsumes other signatures, butresides in a higher dimensional space and contains a lot ofredundant information.
• Identification of landmarks can significantly reduce theredundancies (Boutin)
• It includes the di!erential invariant signature and semi-di!erential invariant signatures as its “coalescent bound-aries”.
• Invariant numerical approximations to di!erential invariantsand semi-di!erential invariants are constructed (usingmoving frames) near these coalescent boundaries.
![Page 56: In v a ria n t H isto g ra m s a n d S ig n a tu re s fo r ...olver/t_/histncsu.pdf · T rian gle in equ ality con e K = { (x ,y,z) | x ,y,z $ 0, x + y $ z, x + z $ y, y + z $ x }](https://reader034.vdocument.in/reader034/viewer/2022050115/5f4c2b7f2acfc309094b10ac/html5/thumbnails/56.jpg)
Statistical Sampling
Idea: Replace high dimensional joint invariant signatures byincreasingly dense point clouds obtained by multiplysampling the original submanifold.
• The equivalence problem requires direct comparison ofsignature point clouds.
• Continuous symmetry detection relies on determining theunderlying dimension of the signature point clouds.
• Discrete symmetry detection relies on determining densities ofthe signature point clouds.