Signed Distance Function Representation, Tracking, and Mapping. Tanner Schmidt

Transcription

1 Signed Distance Function Representation, Tracking, and Mapping Tanner Schmidt

2 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion Patch Volumes DART DynamicFusion

3 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion Patch Volumes DART DynamicFusion

4 Explicit Surface Representations - Geometry is stored explicitly as a list of points, triangles, or other geometric fragments - e.g. meshes, point clouds Vertices: [ (x0, y0, z0), (x1, y1, z1),, (xn, yn, zn) ] Indices: [ (i0, i1), (i2, i3),, (in-1, in) ]

5 Implicit Surface Representation - Geometry is not stored explicitly but rather defined as a level set of a function defined over the space in which the geometry is embedded - There are parametric representations:

6 Implicit Surface Representation - Geometry is not stored explicitly but rather defined as a level set of a function defined over the space in which the geometry is embedded - And there are nonparametric representations:

7 Implicit Surface Representation - Geometry is not stored explicitly but rather defined as a level set of a function defined over the space in which the geometry is embedded - And there are nonparametric representations:

8 Implicit to Explicit Conversion - In two dimensions, we can use an algorithm called marching squares

9 Implicit to Explicit Conversion in 3D - Typically done using marching cubes, a 3D analogue to marching squares

10 Implicit to Explicit Conversion in 3D - Can also be done by raycasting for a view-dependent partial surface

11 Explicit to Implicit Conversion - Can be done by finding the closest point between each discrete location and any part of the geometry

12 Explicit to Implicit Conversion - Can also be done with a distance transform

13 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion PatchVolumes DART DynamicFusion

14 Signed Distance Function Fusion

15 Signed Distance Function Fusion

16 Signed Distance Function Fusion

17 Signed Distance Function Fusion

18 Signed Distance Function Fusion

19 Signed Distance Function Fusion

20 Signed Distance Function Fusion

21 Signed Distance Function Fusion

22 Signed Distance Function Fusion

23 Signed Distance Function Fusion

24 Signed Distance Function Fusion

25 Signed Distance Function Fusion

26 Signed Distance Function Fusion

27 Signed Distance Function Fusion

28 Signed Distance Function Fusion

29 Signed Distance Function Fusion

30 Signed Distance Function Fusion

31 Signed Distance Function Fusion

32 Signed Distance Function Fusion

33 Signed Distance Function Fusion

34 Signed Distance Function Fusion

35 Signed Distance Function Fusion

36 Signed Distance Function Fusion

37 Signed Distance Function Fusion

38 Signed Distance Function Fusion

39 Signed Distance Function Fusion

40 Signed Distance Function Fusion

41 Signed Distance Function Fusion

42 Signed Distance Function Fusion

43 Signed Distance Function Fusion

44 Signed Distance Function Fusion

45 Signed Distance Function Fusion

46 Signed Distance Function Fusion

47 Signed Distance Function Fusion - This addition requires the per-frame projected truncated signed distance volumes to be globally registered

48 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion PatchVolumes DART DynamicFusion

49 Signed Distance Function Tracking

50 Signed Distance Function Tracking

51 Signed Distance Function Tracking

52 Signed Distance Function Tracking

53 Signed Distance Function Tracking

54 Signed Distance Function Tracking

55 Signed Distance Function Tracking

56 Signed Distance Function Tracking

57 Point-plane Iterative Closest Point (ICP)

58 Point-plane Iterative Closest Point (ICP)

59 Point-plane Iterative Closest Point (ICP)

60 Point-plane Iterative Closest Point (ICP)

61 Point-plane Iterative Closest Point (ICP)

62 Point-plane Iterative Closest Point (ICP)

63 Point-plane Iterative Closest Point (ICP)

64 Point-plane Iterative Closest Point (ICP)

65 Direct Signed Distance Function Tracking

66 Direct Signed Distance Function Tracking

67 Direct Signed Distance Function Tracking

68 Online fusion - Tracking requires the fused SDF volume for all frames up to the current frame

69 Online fusion - Tracking requires the fused SDF volume for all frames up to the current frame We must maintain a running average SDF value at each cell

70 Online fusion - Tracking requires the fused SDF volume for all frames up to the current frame We must maintain a running average SDF value at each cell Each cell stores both an SDF value and a weight

71 Truncated Signed Distance Function

72 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion PatchVolumes DART DynamicFusion

73

74 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion PatchVolumes DART DynamicFusion

75 Tracking Failure Color-only Tracking Depth-only Tracking

76 Loop Closure Without Loop Closure With Loop Closure

77

78 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion PatchVolumes DART DynamicFusion

79

80

81

82 Overview - Explicit and implicit surface representations SDF fusion SDF tracking Related research - KinectFusion PatchVolumes DART DynamicFusion

83