Shape Comparison by Size Functions and Natural Size Distances

Shape Comparison by Size Functions and Natural Size Distances

RECENT ADVANCES IN MULTIDIMENSIONAL PERSISTENT TOPOLOGY Patrizio Frosini Vision Mathematics Group University of Bologna - Italy OUTLINE • Introduction and history • Multidimensional size functions • Reduction by the foliation method • Consequences: 1) STABILITY 2) LOCALIZATION OF DISCONTINUITIES • Conclusions Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 2/50 Let us start from three questions... Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 3/50 How similar are the colorings of these leaves? Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 4/50 How similar are the Riemannian structures of these manifolds? Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 5/50 How similar are the spatial positions of these wires? Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 6/50 All previous questions concern the comparison between pairs (M,ϕ), where M is a topological space and ϕ: M → IRk is a continuous function: ●LEAVES: ϕ(P)=color (r,g,b) at P ●3D MODELS: ϕ(P)=vector describing the first fundamental form at P •WIRES: ϕ(P)=(x(P),y(P),z(P)) Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 7/50 Multidimensional Persistent Topology studies the topological properties of the sublevel sets of continuous functions ϕ: M → IRk. It focuses on those properties that “persist” under change of the level. REMARK: the word multidimensional refers to the dimension of IRk, not the dimension of M! Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 8/50 From the beginning of the 90's, the study of Persistent Topology has originated different lines of research. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 9/50 Persistence of connected components (size functions): P. Frosini, A distance for similarity classes of submanifolds of a Euclidean space, Bull. Austral. Math. Soc. 42, 3 (1990), 407-416. A. Verri, C. Uras, P. Frosini, M. Ferri, On the use of size functions for shape analysis, Biological Cybernetics, 70, (1993), 99-107. C. Uras, A. Verri, Computing Size Functions from Edge Maps, International Journal of Computer Vision, 23 (1997), n. 2, 169-183. F. Dibos, P. Frosini and D. Pasquignon, The use of Size Functions for Comparison of Shapes through Differential Invariants, Journal of Mathematical Imaging and Vision, 21, (2004), n. 2, 107-118. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 10/50 Persistence in algebraic topological structures (size homotopy groups, size functor and persistent homology groups): P. Frosini, M. Mulazzani, Size homotopy groups for computation of natural size distances, Bull. Belg. Math. Soc., 6 (1999), 455-464. H. Edelsbrunner, D. Letscher and A. Zomorodian, Topological persistence and simplification, Discrete and Computational Geometry, 28 (2002), 511-533. F. Cagliari, M. Ferri and P. Pozzi, Size Functions from the categorical viewpoint, Acta Applicandae Mathematicae, 67 (2002), 225-235. G. Carlsson, A. Zomorodian, A. Collins and L. Guibas, Persistence barcodes for shapes, International Journal of Shape Modeling, 11(2) (2005), 149-187. D. Cohen-Steiner, H. Edelsbrunner, J. Harer, Stability of persistence diagrams, Discrete and Computational Geometry. 37(1) (2007), 103-120. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 11/50 Natural pseudodistance (metric for persistence comparison) P. Donatini, P. Frosini, Natural pseudodistances between closed manifolds, Forum Mathematicum, 16 (2004), n. 5, 695-715. P. Donatini, P. Frosini, Natural pseudodistances between closed surfaces, Journal of the European Mathematical Society, 9 (2007), 331-353. P. Donatini, P. Frosini, Natural pseudodistances between closed curves, Forum Mathematicum (in press). Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 12/50 The natural pseudodistance Assume that M, N are topological spaces and ϕ : M→IRk, ψ : N→IRk are continuous functions. The natural pseudodistance between the pairs (M,ϕ), (N,ψ ) is defined to be where H(M,N) denotes the set of all homeomorphisms between M and N. Persistent topology can be also seen as a way to obtain lower bounds for the natural pseudodistance. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 13/50 JUST TWO EXAMPLES... Size functions: Theorem. The matching distance between size functions is a lower bound for the natural pseudodistance. Size homotopy groups: Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 14/50 Multidimensional persistence P. Frosini, M. Mulazzani, Size homotopy groups for computation of natural size distances, Bull. Belg. Math. Soc. 6 (1999), 455-464). G. Carlsson, A. Zomorodian, The theory of multidimensional persistence homology, Proc. 2004 Eurographics/ACM SIGGRAPH Symposium on Geometry Processing. ACM Press, New York, NY, 184-193. S. Biasotti, A. Cerri, P. Frosini, D. Giorgi, C. Landi, Multidimensional size functions for shape comparison, Journal of Mathematical Imaging and Vision, 32 (2008), 161-179. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 15/50 Surveys and general papers R. Ghrist, Barcodes: the persistent topology of data, Bull. Amer. Math. Soc., 45(1) (2008), 61-75. S. Biasotti, L. De Floriani, B. Falcidieno, P. Frosini, D. Giorgi, C. Landi, L. Papaleo, M. Spagnuolo, Describing shapes by geometrical-topological properties of real functions, ACM Computing Surveys, 40 (2008), n. 4, 12:1-12:87. H. Edelsbrunner, J.Harer, Persistent homology – a survey, Contemporary Mathematics, 453 (2008), 257-282. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 16/50 Let us go back to size functions Size functions describe the shape of a topological k space M with respect to a function ϕ : Μ → IR . Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 17/50 NOTE: in the following of this talk we ≤ shall write x=(x1,…, xk ) y=(y1,…, yk ) to ≤ mean xi yi for every index i. Analogously, (x1,…, xk ) < (y1,…, yk ) will mean that xi < yi for every index i. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 18/50 Definition of size function Let M be a topological space and assume that a continuous function ϕ : M →IRk is given. For each y ∈ IRk we denote by M 〈ϕ ≤ y〉 the set of all points of M at which the measuring function ϕ takes a value not greater than y. For each real vector y we say that two points P,Q are 〈ϕ ≤ y〉-connected if and only if they 〈ϕ ≤ 〉 belong to the same component in M y . Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 19/50 We call size function (of the pair (M,ϕ)) the ∆+ { ∈ 2k } → function ℓ (M,ϕ ): = (x,y) IR : x < y IN that takes each point (x,y) to the number of equivalence classes of M 〈ϕ ≤ x〉 with respect to 〈ϕ ≤ y〉-connectedness. EQUIVALENT DEFINITION ℓ(M,ϕ )(x,y) is the number of connected components of M〈ϕ ≤ y〉 that contain at least one point of M〈ϕ ≤ x〉. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 20/50 Let us make the definition of size function clear by some example in the case k=1. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 21/50 This is the size function ℓ(M,ϕ ). Example: M is the displayed curve, ϕ is the distance from C. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 22/50 More details about our example Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 23/50 More details about our example Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 24/50 More details about our example Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 25/50 More details about our example Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 26/50 When we move to the multidimensional case (i.e. to the case of measuring functions taking values in IRk instead of IR) the domain of every size function is a subset of IR2k and many new problems arise. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 27/50 ➔How can we represent size functions and persistent homology groups? Which distances can we use to compare size functions (or persistent homology groups)? ➔Are size functions and persistent homology groups stable with respect to small changes of the data? ➔Where are the discontinuities located, in the multidimensional case? Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 28/50 SOME NEW THEORETICAL RESULTS Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 29/50 REPRESENTATION BY FOLIATIONS Introduced in S. Biasotti, A. Cerri, P. Frosini, D. Giorgi, C. Landi, Multidimensional size functions for shape comparison, Journal of Mathematical Imaging and Vision, 32 (2008), 161-179. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 30/50 The theory of k-dimensional size functions can be reduced to the theory of 1-dimensional size functions. Indeed, a parameterized family of half- planes in IRk x IRk exists such that the restriction of ℓ(M,ϕ ) to each of these half- planes can be seen as a particular 1- dimensional size function. Patrizio Frosini (U. Bologna) - MULTIDIMENSIONAL PERSISTENT TOPOLOGY 31/50 k For every unit vector u=(u1,…, uk ) of IR such that ui>0 for i=1,…,k and for every k Σ vector b=(b1,…, bk ) of IR such that i bi=0, we shall say that the pair (u,b) is admissible. Given an admissible pair (u,b), π k k we define the half-plane (u,b) of IR x IR by the following parametric equations: x=su+b s<t {y=tu+b Patrizio Frosini (U.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    50 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us