Presentation is loading. Please wait.

Presentation is loading. Please wait.

Segmentation and Perceptual Grouping. The image of this cube contradicts the optical image.

Similar presentations


Presentation on theme: "Segmentation and Perceptual Grouping. The image of this cube contradicts the optical image."— Presentation transcript:

1 Segmentation and Perceptual Grouping

2

3

4

5

6 The image of this cube contradicts the optical image

7

8

9

10

11

12

13

14

15

16

17 Stochastic Completion Fields

18

19

20 Hough Transform

21

22 Shortest Path

23 Curve Evolution

24

25 Thresholding

26 Histogram

27 Thresholding

28 125 156 99

29

30 Local Uncertainty

31 Global Certainty

32 Local Uncertainty

33 Global Certainty

34 Minimum Cut

35 Texture Examples

36 Coarse Measurements for Texture

37 Hilbert Transform Gaussian Hilbert Transform

38 Filter Bank

39 Textons imagetextons texton assignment

40 Normalized Cuts

41 The Pixel Graph Couplings reflect intensity similarity

42 Hierarchical Graph

43 Normalized-Cut Measure

44 Minimize:

45 Normalized-Cut Measure High-energy cut Low-energy cut

46 Recursive Coarsening

47 Representative subset

48 Recursive Coarsening For a salient segment :, sparse interpolation matrix

49 Weighted Aggregation aggregate

50 Segment Detection

51 A Chicken and Egg Problem Problem: Coarse measurements mix neighboring statistics Solution: support of measurements is determined as the segmentation process proceed

52 Adaptive vs Rigid Support Averaging SWA Geometric

53 Adaptive vs Rigid Support Interpolation SWA Geometric

54 Aggregate Shape

55 Boundary Integrity

56 Sharpen the Aggregates Top-down Sharpening: Expand core Sharpen boundaries

57 Experiments Our SWA algorithm (CVPR’00 + CVPR’01) run-time: 5-10 seconds. Normalized cuts (Shi and Malik, PAMI ’ 00; Malik et al., IJCV ’ 01) run-time: about 10-15 minutes. Software courtesy of Doron Tal, UC Berkeley. images on a pentium III 1000MHz PC:

58 Use of Multiscale Variance Vector

59

60 Variance: Avoid Mixing aggregationMoving window

61 Texture Aggregation Fine (homogeneous) Coarse (heterogeneous)

62 Squirrel

63 Leopard

64 Texture Composition With Variance only With all measures

65 Tiger

66 Butterfly

67 Bird

68 Owl

69 Lion Cub

70 Leopard

71 Polar Bear

72 Segmentation with Ncuts (Malik et al.)

73 Complexity Every level contains about half the nodes of the previous level: Total #nodes double #pixels All connections are local, cleaning small weights Top-down sharpening: constant number of levels Linear complexity Implementation: 5 seconds for 400x400

74

75

76

77

78


Download ppt "Segmentation and Perceptual Grouping. The image of this cube contradicts the optical image."

Similar presentations


Ads by Google