Perceptual grouping: Curvature enhanced closure of elongated structures

30
Perceptual grouping: Curvature enhanced closure of elongated structures By Gijs Huisman Committee: prof. dr. ir. B.M. ter Haar Romeny prof. dr. ir. P. Hilbers dr. L.M.J. Florack dr. ir. R. Duits ir. E.M. Franken

description

Perceptual grouping: Curvature enhanced closure of elongated structures. By Gijs Huisman. Committee: prof. dr. ir. B.M. ter Haar Romeny prof. dr. ir. P. Hilbers dr. L.M.J. Florack dr. ir. R. Duits ir. E.M. Franken. Content. Introduction. Orientation score. - PowerPoint PPT Presentation

Transcript of Perceptual grouping: Curvature enhanced closure of elongated structures

Page 1: Perceptual grouping:  Curvature enhanced closure of elongated structures

Perceptual grouping: Curvature enhanced closure of elongated

structures

ByGijs Huisman

Committee:

prof. dr. ir. B.M. ter Haar Romenyprof. dr. ir. P. Hilbersdr. L.M.J. Florackdr. ir. R. Duitsir. E.M. Franken

Page 2: Perceptual grouping:  Curvature enhanced closure of elongated structures

2/30

Content

1. Introduction

2. Orientation scores• Cake kernels

3. G-convolution• Stochastic completion

kernel• Adaptive G-Convolution

4. Mode line extraction• Theory

5. Non-linear operations• Advection based

enhancement• 3 non-linear operations

6. Curvature estimation• 4 methods• Test results

7. Experiments• Mode line extraction• Increased gap filling• Improved smoothness• Adaptive shooting• Examples medical images

8. Conclusion• Conclusions• Recommendations

Page 3: Perceptual grouping:  Curvature enhanced closure of elongated structures

3/30

Introduction

Page 4: Perceptual grouping:  Curvature enhanced closure of elongated structures

4/30

Orientation score

An orientation score has 2 spatial dimensions and 1 orientation dimension

Page 5: Perceptual grouping:  Curvature enhanced closure of elongated structures

5/30

Orientation Score

An orientation score is obtained by wavelet transformation of an image

Where and

Reconstruction of an image is possible

by an inverse wavelet transform

Page 6: Perceptual grouping:  Curvature enhanced closure of elongated structures

6/30

Orientation scoreCake Kernels

The wavelet is defined by:

The function is defined by B-splines:

Main advantage is easily adaptive kernels with good reconstruction properties

is defined by a 2D gauss

Page 7: Perceptual grouping:  Curvature enhanced closure of elongated structures

7/30

G-convolution

Normal convolution

G-convolution

Page 8: Perceptual grouping:  Curvature enhanced closure of elongated structures

8/30

Stochastic Completion Kernel

G-convolution

The used kernel depicts a probability density function for the continuation of a line kernel in an orientation score.

Page 9: Perceptual grouping:  Curvature enhanced closure of elongated structures

9/30

G-convolutionStochastic Completion Kernel

Gap closing operation with the stochastic completion kernel

Page 10: Perceptual grouping:  Curvature enhanced closure of elongated structures

10/30

G-convolution

Making the G-convolution adaptive means that the kernel properties change with the position in the orientation score.

Kernels are adapted to fit the local curvature

Adaptive

Page 11: Perceptual grouping:  Curvature enhanced closure of elongated structures

11/30

Mode line extraction

Very often the lines itself are demanded instead of an enhanced image.

Any point is part of a local mode line if and at the point

Lines in the spatial plane are 3D ridges in an orientation score.

Page 12: Perceptual grouping:  Curvature enhanced closure of elongated structures

12/30

Non-Linear Operations

Enhancement can be done before and after an G-convolution

Non ideal cake kernel response:

•DC-extraction

•MIN-Extraction

•Erosion

•Advection

Page 13: Perceptual grouping:  Curvature enhanced closure of elongated structures

13/30

Non-Linear Operations

•DC-Extraction

•MIN-Extraction

•Erosion

Page 14: Perceptual grouping:  Curvature enhanced closure of elongated structures

14/30

Non-Linear OperationsAdvection

A force field directed towards the local mode lines:

By means of advection the score can now be sharpened

Page 15: Perceptual grouping:  Curvature enhanced closure of elongated structures

15/30

Non-Linear OperationsResults ErosionDC-extraction

MIN-extraction

Straight

Curved

Advection

Intensity

No preprocessing

Page 16: Perceptual grouping:  Curvature enhanced closure of elongated structures

16/30

Curvature estimation

1. Inner product stochastic completion kernel

2. Inner product Gaussian based kernel

3. Region estimation

4. Hessian based method

Page 17: Perceptual grouping:  Curvature enhanced closure of elongated structures

17/30

Curvature estimationResults

Stochastic Gaussian

Region Hessian

Page 18: Perceptual grouping:  Curvature enhanced closure of elongated structures

18/30

Curvature estimationResults

Curvature measurement on a cross section of the circle line

Stochastic Gaussian

Region Hessian

Page 19: Perceptual grouping:  Curvature enhanced closure of elongated structures

19/30

Curvature estimationResults

Stochastic Gaussian

Region Hessian

Page 20: Perceptual grouping:  Curvature enhanced closure of elongated structures

20/30

ExperimentsMode line extraction

Page 21: Perceptual grouping:  Curvature enhanced closure of elongated structures

21/30

ExperimentsMode line extraction

Page 22: Perceptual grouping:  Curvature enhanced closure of elongated structures

22/30

ExperimentsMode line extraction

Mode line extraction on artificial image

Page 23: Perceptual grouping:  Curvature enhanced closure of elongated structures

23/30

ExperimentsIncreased gap filling

Plane DC Min Plane DC Min

Page 24: Perceptual grouping:  Curvature enhanced closure of elongated structures

24/30

ExperimentsImproved smoothness

Straight

Curved

Page 25: Perceptual grouping:  Curvature enhanced closure of elongated structures

25/30

ExperimentsAdaptive shooting

Original image

Orientation scoreStraight shooting result

Curvature estimate

Enhanced image

Page 26: Perceptual grouping:  Curvature enhanced closure of elongated structures

26/30

ExperimentsAdaptive shooting

Original Straight shooting (1)

Curved Shooting (2)

Curved Shooting (3)

Mean Filling

Method

1.5

1

0.5

02 31

Min Filling

Method

1

2

3

02 31

Page 27: Perceptual grouping:  Curvature enhanced closure of elongated structures

27/30

ExperimentsExamples Medical images

Blood vessel extraction on images of the human retina

Original Threshold Straight shooting

Blood vessels

Page 28: Perceptual grouping:  Curvature enhanced closure of elongated structures

28/30

ExperimentsExamples Medical images

Straight shooting Adaptive shooting

Page 29: Perceptual grouping:  Curvature enhanced closure of elongated structures

29/30

Conclusion

•Curvature enhanced shooting does improve the gap filling

•Successful method of curve extraction

•Good method to estimate the curvature

•Improve the accuracy of the curve extraction method

•Better numerical implementation advection enhancement

•Devise a method to extract the correct curves (e.g. fast marching)

•Better tuning of the cake kernel parameters

Conclusions

Recommendations

Page 30: Perceptual grouping:  Curvature enhanced closure of elongated structures

30/30

Questions?