Presentation is loading. Please wait.

Presentation is loading. Please wait.

A topological approach to watersheds

Similar presentations


Presentation on theme: "A topological approach to watersheds"— Presentation transcript:

1 A topological approach to watersheds
G. Bertrand Laboratoire A2SI, ESIEE Paris 27/05/2003

2 To our knowledge there is no framework in which general properties for watersheds can be derived
Our goal is to show that the topological approach proposed previously* provides such a framework *M. Couprie and G. Bertrand (1997) Paris 27/05/2003

3 Watersheds Powerful segmentation operator from the field of Mathematical Morphology Introduced as a tool for segmenting grayscale images by S. Beucher, H. Digabel and C. Lantuejoul in the 70s Efficient algorithms based on flooding simulation were proposed by F. Meyer, P. Soille, L. Vincent (and others) in the 90s Paris 27/05/2003

4 Flooding paradigm Paris 27/05/2003

5 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 1 20 40 40 3 3 5 5 30 30 30 10 15 1 15 1 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 1 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 1 15 10 1 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

6 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 20 1 40 40 3 3 5 5 30 30 30 10 15 1 15 20 1 40 40 3 3 5 30 20 20 20 30 15 1 15 1 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 1 10 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 15 1 20 15 1 10 5 1 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 There is no descending path from the 20s to the minimum 3 The contrast between minima is not preserved Paris 27/05/2003

7 Discrete sets and destructible points
Let G = (V,E) be a (undirected) graph and let X be a subset of V. We say that a point x  X is destructible for X if x is adjacent to exactly one connected component of X. M. Couprie and G. Bertrand (1997) Paris 27/05/2003

8 Destructible point Paris 27/05/2003

9 Destructible point Paris 27/05/2003

10 Destructible point Paris 27/05/2003

11 Destructible point Paris 27/05/2003

12 Destructible point Paris 27/05/2003

13 Destructible point Paris 27/05/2003

14 Destructible point Paris 27/05/2003

15 Destructible point Paris 27/05/2003

16 Destructible point Paris 27/05/2003

17 Destructible point Paris 27/05/2003

18 Discrete maps and destructible points
Let G = (V,E) be a connected (undirected) graph. We denote by F (V) the family composed of all maps from V to Z. Let F  F (V), we set Fk = {x  V; F(x)  k}, Fk is the cross-section of F at level k Let x  V and let k = F(x). We say that x is destructible (for F) if x is adjacent to exactly one connected component of Fk M. Couprie and G. Bertrand (1997) Paris 27/05/2003

19 Topological watershed
Let F and F’ be in F (V). We say that F’ is a thinning of F if F’ may be obtained from F by iteratively lowering destructible points (by 1). Let F and F’ be in F (V). We say that F’ is a watershed of F if F’ is a thinning of F and if there is no destructible point for F’. M. Couprie and G. Bertrand (1997) Paris 27/05/2003

20 Collapsing paradigm Paris 27/05/2003

21 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 1 20 40 40 3 3 5 5 30 30 30 10 15 1 15 1 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 1 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 1 15 10 1 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

22 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 1 20 30 3 3 3 40 40 30 30 30 1 1 20 30 30 30 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 The watershed is located on the crest lines of the original image The contrast between minima is preserved Paris 27/05/2003

23 Pass value Let F be in F (V). If п is a path, we set F(п) = Max{F(x); x  п}. Let x, y in V. We set F(x,y) = Min {F(п); п  п(x,y)}, F(x,y) is the pass value between x and y. Let X and Y be two subsets of V. We set F(X,Y) = Min{F(x,y); x  X and y  Y}. Paris 27/05/2003

24 Separation Let F be in F (V) and let x and y be in V. We say that x and y are separated (for F) if F(x,y) > Max{F(x),F(y)}. We say that x and y are k-separated (for F) if x and y are separated and F(x,y) = k. Let F and F’ be in F (V) such that F’  F. We say that F’ is a separation of F if, for all x,y in V, if x and y are k-separated for F, then x and y are k-separated for F’. Paris 27/05/2003

25 k-separation y x x and y are 20-separated 40 40 40 40 40 40 40 40 40
1 1 2 3 10 5 25 5 4 4 4 40 x y 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

26 x and y are NOT separated (they are linked)
k-separation x and y are NOT separated (they are linked) 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 3 10 5 25 5 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 x y 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

27 Theorem (restriction to minima)
Let F and F’ be in F (V) such that F’  F. The map F’ is a separation of F if and only if, for all distinct minima X,Y for F, F(X,Y) = F’(X,Y). Paris 27/05/2003

28 Theorem (strong separation)
Let F and F’ be in F (V) such that F’  F. The map F’ is a thinning of F if and only if F’ is a strong separation of F. Paris 27/05/2003

29 2D case Any connected object without hole reduces to one point
Paris 27/05/2003

30 Bing’s house with two rooms
3D case Some connected objects without holes and cavities DO NOT reduce to one point Bing’s house with two rooms Paris 27/05/2003

31 Theorem (restriction to minima)
Let F and F’ be in F (V) such that F’  F. The map F’ is a separation of F if and only if, for all distinct minima X,Y for F, F(X,Y) = F’(X,Y). Paris 27/05/2003

32 Theorem (restriction to minima)
Let F and F’ be in F (V) such that F’  F. The map F’ is a separation of F if and only if, for all distinct minima X,Y for F, F(X,Y) = F’(X,Y). Is it possible to reduce the amount of information necessary to encode the  "topology" of a thinning? Paris 27/05/2003

33 Ordered minima Let F be in F (V). A minima ordering (for F) is a strict total order relation < on the minima of F. Let X be a minimum for F. The pass value of X for (F,<) is the number F(X,<) such that: i) if X = Xmin, then F(X,<) = infinity; ii) otherwise, F(X,<) = Min {F(X,Y); for all minima Y such that Y < X}. Paris 27/05/2003

34 Ordered minima F(.,<)=8 5 3 2 F(.,<)=22 F(.,<)=30 1 4
40 40 40 40 40 40 40 40 40 40 40 40 40 5 40 3 1 1 2 3 10 5 25 5 4 4 4 40 2 40 1 2 8 6 5 5 20 3 2 3 40 F(.,<)=22 40 3 3 2 3 10 6 6 6 22 2 3 F(.,<)=30 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 1 4 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 F(.,<)=infty F(.,<)=31 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

35 Theorem (ordered minima)
Let F and F’ be in F (V) such that F’ <= F and let < be a minima ordering for F. The map F’ is a separation of F if and only if, for each minimum X for F, we have F(X,<) = F’(X,<). Paris 27/05/2003

36 <-map F(.,<)=8 5 3 2 F(.,<)=22 F(.,<)=30 1 4 F(.,<)=0
40 40 40 40 40 40 40 40 40 40 40 40 40 5 40 3 1 1 2 3 10 5 25 5 4 4 4 40 2 40 1 2 8 6 5 5 20 3 2 3 40 F(.,<)=22 40 3 3 2 3 10 6 6 6 22 2 3 F(.,<)=30 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 1 4 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 F(.,<)=0 F(.,<)=31 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

37 Theorem (reconstruction from ordered pass values)
Let F be in F (V) and let < be a minima ordering for F. Let T be a tree associated to F(X,<). The pass values between all minima of F may be reconstructed from T. Paris 27/05/2003

38 Dynamics (M. Grimaud,1992) Let X be a minimum for F Let G(X) be the number such that: i) if X = Xmin, then G(X) = infinity; ii) otherwise, G(X) = Min {F(X,Y); for all minima Y such that F(Y) < F(X)}. The dynamics of a minimum X is the number Dyn(X) = G(X) – F(X) Paris 27/05/2003

39 Ordered dynamics The notion of ordered pass values leads to a new definition of the dynamics of a minimum: Dyn(X; F, <) = F(X, <) – F(X) This new definition of dynamics fully agrees with the notion of separation. Paris 27/05/2003

40 Dynamics: counter-example
Paris 27/05/2003

41 Dynamics: counter-example
Paris 27/05/2003

42 Ordered minima 2 3 1 Paris 27/05/2003

43 Conclusion Basins, component tree Duality: minimum spanning trees
Comparison of existing algorithms (L. Najman and M. Couprie) Algorithmic issues (M. Couprie and L. Najman) Saliency (L. Najman) Paris 27/05/2003

44 Results of watershed algorithms
4 5 B 6 3 1 C 2 Topographical 3 4 5 6 2 1 Original image A B 6 1 C 3 5 Vincent-Soille, Meyer and Topological Paris 27/05/2003

45 Results of watershed algorithms
30 3 31 4 255 2 1 5 30 C 31 D E A 255 B F Original image Vincent-Soille 30 C E D A 255 B F 30 C 31 D 255 A E B F Meyer Topological Paris 27/05/2003

46 Results of watershed algorithms
2 1 30 20 40 A B 1 C 20 40 Original image Vincent-Soille A B 1 20 40 A A A A A A 1 A A A 30 30 30 A A 1 A 30 B 20 A 30 A 1 40 B B 20 A A 40 B B B 20 A A 1 A B B B 20 A A A B B 1 B 20 1 A A 1 A Paris 27/05/2003 Meyer Topological

47 Homotopy: an illustration
F(x,y) G(x,y) F1 G1 x x Paris 27/05/2003

48 Homotopy: an illustration
F(x,y) G(x,y) x x F2 G2 F1 G1 Paris 27/05/2003

49 Watershed transform Paris 27/05/2003

50 Flooding paradigm Paris 27/05/2003

51 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 1 20 40 40 3 3 5 5 30 30 30 10 15 1 15 1 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 1 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 1 15 10 1 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

52 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 1 15 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 5 1 5 1 40 40 1 3 5 10 15 1 20 1 15 10 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

53 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

54 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

55 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

56 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

57 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

58 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

59 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

60 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

61 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 1 10 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 15 1 10 1 5 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

62 Flooding paradigm 40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 15 1 20 40 40 3 3 5 5 30 30 30 10 1 15 15 20 40 40 3 3 5 30 20 20 20 30 15 1 1 15 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 15 1 20 1 15 10 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 There is no descending path from the 20s to the minimum 3 Any path from 0 to the minimum 3 must climb at least at 30 Paris 27/05/2003

63 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 1 20 40 40 3 3 5 5 30 30 30 10 15 1 15 1 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 1 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 1 15 10 1 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

64 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 1 20 40 40 3 3 5 5 30 30 30 10 15 1 15 1 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 1 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 1 15 10 1 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

65 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 1 20 40 40 3 3 5 5 30 30 30 10 15 1 15 1 20 40 40 3 3 5 30 20 20 20 30 15 1 15 1 20 1 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 10 1 10 1 10 1 40 40 5 5 5 10 40 20 40 10 10 1 5 5 1 40 40 1 3 5 10 1 15 20 1 15 10 1 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 1 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

66 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 20 1 40 40 3 3 5 5 30 30 30 10 15 1 15 20 1 40 40 3 3 5 30 20 20 20 30 15 1 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 1 10 10 1 40 40 5 5 5 10 40 20 40 10 10 5 1 1 5 40 40 1 3 5 10 1 15 20 1 15 10 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

67 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 5 5 5 10 10 10 1 10 1 15 20 1 40 40 3 3 5 5 30 30 30 10 15 1 15 20 1 40 40 3 3 5 30 20 20 20 30 15 1 1 15 1 20 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 10 10 10 40 20 20 20 40 1 10 1 10 10 1 40 40 5 5 5 10 40 20 40 10 10 5 1 1 5 40 40 1 3 5 10 1 15 20 1 15 10 5 1 1 1 40 40 1 3 5 10 15 20 15 10 5 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

68 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

69 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

70 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

71 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

72 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

73 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

74 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 40 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

75 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 1 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

76 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 1 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

77 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 1 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

78 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

79 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 20 20 20 30 3 3 3 40 40 40 40 40 40 20 20 20 40 40 40 40 40 40 1 1 1 40 20 20 20 40 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

80 Topological watershed
40 40 40 40 40 40 40 40 40 40 40 40 40 40 3 3 3 3 3 3 3 3 3 3 3 40 40 3 3 3 3 30 30 30 3 3 3 3 40 40 3 3 3 30 1 20 30 3 3 3 40 40 30 30 30 1 1 20 30 30 30 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 1 1 1 1 1 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 The watershed is located on the crest lines of the original image The contrast between minima is preserved Paris 27/05/2003

81 k-separation y x x and y are 8-separated 40 40 40 40 40 40 40 40 40 40
1 1 2 3 10 5 25 5 4 4 4 40 x y 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

82 x and y are NOT separated (they are linked)
k-separation x and y are NOT separated (they are linked) 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 3 10 5 25 5 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 x y 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

83 Homotopic thinning Paris 27/05/2003

84 Homotopic thinning Paris 27/05/2003

85 Homotopic thinning Paris 27/05/2003

86 Homotopic thinning Paris 27/05/2003

87 Homotopic thinning Paris 27/05/2003

88 Homotopic thinning Paris 27/05/2003

89 Homotopic thinning Paris 27/05/2003

90 Homotopic thinning Paris 27/05/2003

91 Homotopic thinning Paris 27/05/2003

92 Pass value 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 3 10 5 25 5 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

93 Pass value 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 3 10 5 25 5 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

94 Pass value 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 3 10 5 25 5 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003

95 Pass value 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 3 10 5 25 5 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 40 6 6 40 6 11 11 11 25 4 4 4 40 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 F(X,Y) = 31 Paris 27/05/2003

96 Pass value 20 40 40 40 40 40 40 40 40 40 40 40 40 40 40 1 1 2 8 3 10 5 25 5 20 4 4 4 40 40 1 2 8 6 5 5 20 3 2 3 40 40 3 3 2 3 10 6 6 6 22 2 3 40 30 31 40 6 6 40 6 11 11 11 25 4 4 4 40 31 30 40 40 35 10 30 15 15 15 35 31 36 10 40 40 10 8 5 10 32 33 34 10 10 15 38 40 40 8 5 1 1 15 40 10 6 3 15 20 40 40 10 8 5 10 15 35 15 6 6 15 35 40 40 40 40 40 40 40 40 40 40 40 40 40 40 Paris 27/05/2003 31


Download ppt "A topological approach to watersheds"

Similar presentations


Ads by Google