an efficient algorithm for determining an aesthetic shape
play

An Efficient Algorithm for Determining an Aesthetic Shape - PowerPoint PPT Presentation

An Efficient Algorithm for Determining an Aesthetic Shape Connecting Unorganised 2D Points Stefan Ohrhallinger 1,2 and Sudhir Mudur 2 1 Vienna University of Technology, 2 Concordia University, Montral Connect The Dots Stefan Ohrhallinger and


  1. Start With A Minimum Set NP-hard P MST Vertex degree Stefan Ohrhallinger and Sudhir Mudur 49

  2. Start With A Minimum Set NP-hard P MST Vertex degree Stefan Ohrhallinger and Sudhir Mudur 50

  3. Start With A Minimum Set NP-hard? NP-hard P MST Vertex degree Stefan Ohrhallinger and Sudhir Mudur 51

  4. Start With A Minimum Set NP-hard P MST Vertex degree Stefan Ohrhallinger and Sudhir Mudur 52

  5. Start With A Minimum Set NP-hard ? P MST Vertex degree Stefan Ohrhallinger and Sudhir Mudur 53

  6. Overview Of Our 'Connect2D' Algorithm Input Stefan Ohrhallinger and Sudhir Mudur 54

  7. Overview Of Our 'Connect2D' Algorithm Input Stefan Ohrhallinger and Sudhir Mudur 55

  8. Overview Of Our 'Connect2D' Algorithm Input Stefan Ohrhallinger and Sudhir Mudur 56

  9. Overview Of Our 'Connect2D' Algorithm not manifold Input Stefan Ohrhallinger and Sudhir Mudur 57

  10. Overview Of Our 'Connect2D' Algorithm Inflate min Δ|B| Input Stefan Ohrhallinger and Sudhir Mudur 58

  11. Overview Of Our 'Connect2D' Algorithm Inflate min Δ|B| Input Stefan Ohrhallinger and Sudhir Mudur 59

  12. Overview Of Our 'Connect2D' Algorithm Inflate Sculpture min Δ|B| min Δ|B| Input Stefan Ohrhallinger and Sudhir Mudur 60

  13. Overview Of Our 'Connect2D' Algorithm Inflate Dual Sculpture min Δ|B| min Δ|B| Input Stefan Ohrhallinger and Sudhir Mudur 61

  14. Overview Of Our 'Connect2D' Algorithm Inflate Dual Sculpture min Δ|B| min Δ|B| Input Theorem 1 : Our algorithm constructs B out in O(n log n) time. Stefan Ohrhallinger and Sudhir Mudur 62

  15. Improved For Sparse Sampling Points Stefan Ohrhallinger and Sudhir Mudur 63

  16. Improved For Sparse Sampling Points Gathan [DW01] Stefan Ohrhallinger and Sudhir Mudur 64

  17. Improved For Sparse Sampling Points Gathan [DW01] Ours [OM13] Stefan Ohrhallinger and Sudhir Mudur 65

  18. Large Sets: Easy 10k points Gathan [DW01] Stefan Ohrhallinger and Sudhir Mudur 66

  19. Large Sets: Easy 10k points Gathan [DW01] Ours [OM13]: manifold Stefan Ohrhallinger and Sudhir Mudur 67

  20. Challenge: Extremely Sparse Sampling Points Stefan Ohrhallinger and Sudhir Mudur 68

  21. Challenge: Extremely Sparse Sampling Points Gathan [DW01] Stefan Ohrhallinger and Sudhir Mudur 69

  22. Challenge: Extremely Sparse Sampling Points Gathan [DW01] Ours Stefan Ohrhallinger and Sudhir Mudur 70

  23. Failure Cases [OM11] Stefan Ohrhallinger and Sudhir Mudur 71

  24. Failure Cases local minimum [OM11] Ours [OM13] Stefan Ohrhallinger and Sudhir Mudur 72

  25. Not Bmin? Insert Points, Manually Not a solid Stefan Ohrhallinger and Sudhir Mudur 73

  26. Not Bmin? Insert Points, Manually local minimum Not a solid Stefan Ohrhallinger and Sudhir Mudur 74

  27. Not Bmin? Insert Points, Manually local minimum undersampled Not a solid Stefan Ohrhallinger and Sudhir Mudur 75

  28. Not Bmin? Insert Points, Manually local minimum + undersampled Not a solid Stefan Ohrhallinger and Sudhir Mudur 76

  29. Not Bmin? Insert Points, Manually local minimum + + undersampled Not a solid Stefan Ohrhallinger and Sudhir Mudur 77

  30. Robust To Noise [MTSM10] Stefan Ohrhallinger and Sudhir Mudur 78

  31. Robust To Noise [MTSM10] Ours [OM13]: manifold+interpolating Stefan Ohrhallinger and Sudhir Mudur 79

  32. Class Of Point Sets C Stefan Ohrhallinger and Sudhir Mudur 80

  33. Class Of Point Sets C M Medial axis M Stefan Ohrhallinger and Sudhir Mudur 81

  34. Class Of Point Sets C M Medial axis M Local feature size f(x)=||x, M|| Stefan Ohrhallinger and Sudhir Mudur 82

  35. Class Of Point Sets C M Medial axis M Local feature size f(x)=||x, M|| Stefan Ohrhallinger and Sudhir Mudur 83

  36. Class Of Point Sets C M Medial axis M Local feature size f(x)=||x, M|| ε -sampling: Stefan Ohrhallinger and Sudhir Mudur 84

  37. Class Of Point Sets C M Medial axis M Local feature size f(x)=||x, M|| ε -sampling: Stefan Ohrhallinger and Sudhir Mudur 85

  38. Class Of Point Sets C M Medial axis M Local feature size f(x)=||x, M|| ε -sampling: Stefan Ohrhallinger and Sudhir Mudur 86

  39. Class Of Point Sets C M Theorem 2 BC 0 reconstructs ε -sampled C from P with ε < 0.5 and a local non-uniformity u < 1.609. Stefan Ohrhallinger and Sudhir Mudur 87

  40. Class Of Point Sets C M Theorem 2 BC 0 reconstructs ε -sampled C from P with ε < 0.5 and a local non-uniformity u < 1.609. Stefan Ohrhallinger and Sudhir Mudur 88

  41. Class Of Point Sets C M Theorem 2 [ABE98] 1 BC 0 reconstructs ε -sampled C from P with ε < 0.5 and a local non-uniformity u < 1.609. 1 please see paper for references. Stefan Ohrhallinger and Sudhir Mudur 89

  42. Class Of Point Sets C M Theorem 2 [ABE98] 1 BC 0 reconstructs ε -sampled C from P with [DK99] 1 ε < 0.5 and a local non-uniformity u < 1.609. 1 please see paper for references. Stefan Ohrhallinger and Sudhir Mudur 90

  43. Class Of Point Sets C M Theorem 2 [ABE98] 1 BC 0 reconstructs ε -sampled C from P with [DK99] 1 ε < 0.5 and a local non-uniformity u < 1.609. BC 0 =B min 1 please see paper for references. Stefan Ohrhallinger and Sudhir Mudur 91

  44. Class Of Point Sets C M Theorem 2 [ABE98] 1 BC 0 reconstructs ε -sampled C from P with B=B min [DK99] 1 ε < 0.5 and a local non-uniformity u < 1.609. (conj) BC 0 =B min 1 please see paper for references. Stefan Ohrhallinger and Sudhir Mudur 92

  45. Why Limited Non-Uniformity? Stefan Ohrhallinger and Sudhir Mudur 93

  46. Why Limited Non-Uniformity? Stefan Ohrhallinger and Sudhir Mudur 94

  47. Future Work (2D) Local minimum → fill hole Stefan Ohrhallinger and Sudhir Mudur 95

  48. Future Work (2D) Local minimum → fill hole Multiply connected components Stefan Ohrhallinger and Sudhir Mudur 96

  49. Future Work (2D) Prove B min for ε < 1 Local minimum → fill hole Multiply connected components Stefan Ohrhallinger and Sudhir Mudur 97

  50. Future Work (2D) Prove B min for ε < 1 Local minimum → fill hole ε < 0.5 Sampling: tighter bound Multiply connected components Stefan Ohrhallinger and Sudhir Mudur 98

  51. Future Work (2D) Prove B min for ε < 1 Local minimum → fill hole ε < 0.5 Sampling: tighter bound Multiply connected components Open curves (vs. sparse) Stefan Ohrhallinger and Sudhir Mudur 99

  52. Future Work (3D) See my talk at SMI'13 (July 11th), Bournemouth, UK Stefan Ohrhallinger and Sudhir Mudur 100

Recommend


More recommend