Presentation is loading. Please wait.

Presentation is loading. Please wait.

Segmentation and Boundary Detection Using Multiscale Intensity Measurements Eitan Sharon, Meirav Galun, Ronen Basri, Achi Brandt Dept. of Computer Science.

Similar presentations


Presentation on theme: "Segmentation and Boundary Detection Using Multiscale Intensity Measurements Eitan Sharon, Meirav Galun, Ronen Basri, Achi Brandt Dept. of Computer Science."— Presentation transcript:

1 Segmentation and Boundary Detection Using Multiscale Intensity Measurements Eitan Sharon, Meirav Galun, Ronen Basri, Achi Brandt Dept. of Computer Science and Applied Mathematics The Weizmann Institute of Science

2 Eitan Sharon - Weizmann Institute Image Segmentation

3 Eitan Sharon - Weizmann Institute Local Uncertainty

4 Eitan Sharon - Weizmann Institute Global Certainty

5 Eitan Sharon - Weizmann Institute Local Uncertainty

6 Eitan Sharon - Weizmann Institute Global Certainty

7 Eitan Sharon - Weizmann Institute Coarse Measurements for Texture

8 Eitan Sharon - Weizmann Institute A Chicken and Egg Problem Problem: Coarse measurements mix neighboring statistics Solution: support of measurements is determined as the segmentation process proceeds

9 Eitan Sharon - Weizmann Institute  Normalized-cuts measure in graphs  Complete hierarchy in linear time  Use multiscale measures of intensity, texture, shape, and boundary integrity Segmentation by Weighted Aggregation

10 Eitan Sharon - Weizmann Institute  Normalized-cuts measure in graphs  Complete hierarchy in linear time  Use multiscale measures of intensity, texture, shape, and boundary integrity Segmentation by Weighted Aggregation

11 Eitan Sharon - Weizmann Institute Segmentation by Weighted Aggregation  Normalized-cuts measure in graphs  Complete hierarchy in linear time  Use multiscale measures of intensity, texture, shape and boundary integrity

12 Eitan Sharon - Weizmann Institute The Pixel Graph Couplings Reflect intensity similarity Low contrast – strong coupling High contrast – weak coupling

13 Eitan Sharon - Weizmann Institute Hierarchical Graph

14 Eitan Sharon - Weizmann Institute Hierarchy in SWA

15 Eitan Sharon - Weizmann Institute Normalized-Cut Measure

16 Eitan Sharon - Weizmann Institute Normalized-Cut Measure

17 Eitan Sharon - Weizmann Institute Normalized-Cut Measure

18 Eitan Sharon - Weizmann Institute Normalized-Cut Measure Minimize:

19 Eitan Sharon - Weizmann Institute Normalized-Cut Measure High-energy cut Minimize:

20 Eitan Sharon - Weizmann Institute Normalized-Cut Measure Low-energy cut Minimize:

21 Eitan Sharon - Weizmann Institute Recursive Coarsening

22 Eitan Sharon - Weizmann Institute Recursive Coarsening Representative subset

23 Eitan Sharon - Weizmann Institute Recursive Coarsening For a salient segment :, sparse interpolation matrix

24 Eitan Sharon - Weizmann Institute Weighted Aggregation aggregate

25 Eitan Sharon - Weizmann Institute Segment Detection

26 Eitan Sharon - Weizmann Institute SWA Linear in # of points (a few dozen operations per point) Detects the salient segments Hierarchical structure

27 Eitan Sharon - Weizmann Institute Coarse-Scale Measurements Average intensities of aggregates Multiscale intensity-variances of aggregates Multiscale shape-moments of aggregates Boundary alignment between aggregates

28 Eitan Sharon - Weizmann Institute Adaptive vs. Rigid Measurements Averaging Our algorithm - SWA Geometric Original

29 Eitan Sharon - Weizmann Institute Our algorithm - SWA Adaptive vs. Rigid Measurements Interpolation Geometric Original

30 Eitan Sharon - Weizmann Institute Recursive Measurements: Intensity aggregate intensity of pixel i average intensity of aggregate

31 Eitan Sharon - Weizmann Institute Use Averages to Modify the Graph

32 Eitan Sharon - Weizmann Institute Use Averages to Modify the Graph

33 Eitan Sharon - Weizmann Institute Texture Examples

34 Eitan Sharon - Weizmann Institute Isotropic and Oriented Filters Textons by K-Means Malik et al IJCV2001 A brief tutorial

35 Eitan Sharon - Weizmann Institute Isotropic Texture in SWA Intensity Variance Isotropic Texture of aggregate – average of variances in all scales

36 Eitan Sharon - Weizmann Institute Isotropic Texture in SWA Intensity Variance Isotropic Texture of aggregate – average of variances in all scales

37 Eitan Sharon - Weizmann Institute Isotropic Texture in SWA Intensity Variance Isotropic Texture of aggregate – average of variances in all scales

38 Eitan Sharon - Weizmann Institute Oriented Texture in SWA Shape Moments Oriented Texture of aggregate – orientation, width and length in all scales center of mass width length orientation with Meirav Galun

39 Eitan Sharon - Weizmann Institute Gestalt – Perceptual Grouping A brief Tutorial Sharon, Brandt, Basri PAMI 2000: Shashua and Ullman ICCV 1988: Group curves by: Proximity Co-linearity

40 Eitan Sharon - Weizmann Institute Boundary Integrity in SWA

41 Eitan Sharon - Weizmann Institute Sharpen the Aggregates Top-down Sharpening: Expand core Sharpen boundaries

42 Eitan Sharon - Weizmann Institute Hierarchy in SWA

43 Eitan Sharon - Weizmann Institute 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:

44 Eitan Sharon - Weizmann Institute Isotropic Texture - Horse I Our Algorithm (SWA) Normalized Cuts

45 Eitan Sharon - Weizmann Institute Isotropic Texture - Horse II Our Algorithm (SWA)Normalized Cuts

46 Eitan Sharon - Weizmann Institute Isotropic Texture - Tiger Normalized Cuts Our Algorithm (SWA)

47 Eitan Sharon - Weizmann Institute Isotropic Texture - Butterfly Our Algorithm (SWA) Normalized Cuts

48 Eitan Sharon - Weizmann Institute Isotropic Texture - Leopard Our Algorithm (SWA)

49 Eitan Sharon - Weizmann Institute Isotropic Texture - Dalmatian Dog Our Algorithm (SWA)

50 Eitan Sharon - Weizmann Institute Isotropic Texture - Squirrel Our Algorithm (SWA)Normalized Cuts

51 Eitan Sharon - Weizmann Institute Full Texture - Squirrel Our Algorithm (SWA)Normalized Cuts with Meirav Galun

52 Eitan Sharon - Weizmann Institute Full Texture - Composition Our Algorithm (SWA) with Meirav Galun

53 Eitan Sharon - Weizmann Institute Full Texture – Lion Cub Our Algorithm (SWA) with Meirav Galun

54 Eitan Sharon - Weizmann Institute Full Texture – Polar Bear Our Algorithm (SWA) with Meirav Galun

55 Eitan Sharon - Weizmann Institute Full Texture – Penguin Our Algorithm (SWA) with Meirav Galun

56 Eitan Sharon - Weizmann Institute Full Texture –Leopard Our Algorithm (SWA) with Meirav Galun

57 Eitan Sharon - Weizmann Institute Full Texture - Zebra Our Algorithm (SWA) with Meirav Galun

58 Eitan Sharon - Weizmann Institute Segmentation by Weighted Aggregation Efficient approximation to Ncut-like measures Recursive computation of multiscale measurements Novel adaptive pyramid representing the image

59 Eitan Sharon - Weizmann Institute Matching Experiment (Chen Brestel)

60 Eitan Sharon - Weizmann Institute Matching Experiment (Chen Brestel)

61 Eitan Sharon - Weizmann Institute Matching Experiment (Chen Brestel)

62 Eitan Sharon - Weizmann Institute Experiments – Clustering Silhouettes Data from: Gdalyahu, Weinshall, Werman – CVPR 99 (Also: Domany, Blatt, Gdalyahu, Weinshall)

63 Eitan Sharon - Weizmann Institute Descending order of prominence – objects are fully categorized most prominentless prominent

64 Eitan Sharon - Weizmann Institute Bottom-Up and Top-Down


Download ppt "Segmentation and Boundary Detection Using Multiscale Intensity Measurements Eitan Sharon, Meirav Galun, Ronen Basri, Achi Brandt Dept. of Computer Science."

Similar presentations


Ads by Google