geometric (bio-) modeling and visualization

26
Geometric (Bio-) Modeling and Visualization http://www.cs.utexas.edu/~bajaj/c3s84R10/ Lecture 19b Maps IIb: Image/Surface - Filtering

Upload: others

Post on 28-Apr-2022

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Geometric (Bio-) Modeling and Visualization

Geometric (Bio-) Modeling and Visualizationhttp://www.cs.utexas.edu/~bajaj/c3s84R10/

Lecture 19b

Maps IIb: Image/Surface - Filtering

Page 2: Geometric (Bio-) Modeling and Visualization

2010!"#$"%&'(%&!()*+$,$-(#,.&/-0+,.-1,$-(#

2#0$-$+$"&('&!()*+$,$-(#,.&,#3&4#5-#""%-#5&67-"#7"0

8"*,%$)"#$&('&!()*+$"%&67-"#7"0&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&9#-:"%0-$;&('&<"=,0&,$&>+0$-# !"#$"%&"'(()**7

6+?@#,#()"$"%&6$%+7$+%"&4.+7-3,$-(#&'%()&A8&!%;(@4B

!"#$%&'(!))*(+,-./(01(2.3$4-5"6357$4((!847-$5"$973(,4/(:.35$"(17;;6-7$4()7<5."74=(!>5"6356".(>.=?.45,57$4(!@6,-7%85$?73('$/.<74=(((!:7-6,<7A,57$4(

!"#$$%&#'%(#')*+,%- "-./01"230+4%5$#'+"#$$676+#8+269.:.;60+<;9'6= >%$.0+?">@A

++!"#$%#&'()*+),!-./012),/3

!"#$%&'()*%+"),-%./0123

2010

Page 3: Geometric (Bio-) Modeling and Visualization

2010!"#$"%&'(%&!()*+$,$-(#,.&/-0+,.-1,$-(#

2#0$-$+$"&('&!()*+$,$-(#,.&,#3&4#5-#""%-#5&67-"#7"0

8"*,%$)"#$&('&!()*+$"%&67-"#7"0&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&9#-:"%0-$;&('&<"=,0&,$&>+0$-# !"#$"%&"'(()**7

!"#$%&'$&%(")($(%*+,-$+.,"/+0(1+,("2.%"3+,41("0-%$+'1("'%5.678

Page 4: Geometric (Bio-) Modeling and Visualization

Center for Computational Visualization http://www.ices.utexas.edu/CCVInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin May 2008

Bilateral & Anisotropic FilteringBilateral filtering

where and are parameters

and f(.) is the image intensity value.

Anisotropic diffusion filtering

where a stands for the diffusion tensor

determined by local curvature estimation.W. Jiang, M. Baker, Q. Wu, C. Bajaj, W. Chiu, Journal of Structural

Biology, 144, 5,(2003), Pages 114-122

C. Bajaj, G. Xu, ACM Transactions on Graphics, (2003),22(1), pp. 4- 32.

Page 5: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Image Denoising (Filtering): A Variational Approach

Page 6: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 7: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 8: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 9: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 10: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 11: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 12: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Page 13: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Derivative Relationships

H(x) is the Mean Surface Curvature

Page 14: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Anisotropic Diffusion Filtering

Page 15: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

be principal curvature directions of point

Shape Modulation via Diffusion Tensor

Page 16: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Functions on Surface: Texture

Initial data After 1 iteration After 4 iterations

Page 17: Geometric (Bio-) Modeling and Visualization

Center for Computational Visualization http://www.ices.utexas.edu/CCVInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin May 2008

where is a decreasing function is the angle between the central pixel and its surrounding pixels.

• Gradient vector diffusion:- smoothing the vector fields

- diffusion to flat regions

• For smooth data:- zeroes of the gradient vector field

- simple, easy to implement

• For noisy data:- Gradient vector diffusion

- higher time complexity but robust to noise

Filtering Gradient Map Critical Points

minimum maximum saddle (0) (3) (1, 2)

Y.Zu, C. Bajaj IEEE Transactions on Image Processin, 2005, 14, 9, 1324-1337

Page 18: Geometric (Bio-) Modeling and Visualization

2010!"#$"%&'(%&!()*+$,$-(#,.&/-0+,.-1,$-(#

2#0$-$+$"&('&!()*+$,$-(#,.&,#3&4#5-#""%-#5&67-"#7"0

8"*,%$)"#$&('&!()*+$"%&67-"#7"0&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&9#-:"%0-$;&('&<"=,0&,$&>+0$-# !"#$"%&"'(()**7

>#-0($%(*-7&?%,3-"#$&/"7$(%&8-''+0-(#&

@>?/8A

!!"

!!#

$

%&&'())

%&&'())

++,,

++,,

)))

)))

!"!

!""

###$$%

$

###&&%

&

*

*

-."'"/(((

,0,$+1(2,$++(3$4563(78'($."("2892:5;(2"<$8'(7:"96=

* :3(4(<853$45$=(

#':3($."(8':;:549(:%4;"($8(&"(6:7703"6=(

,#" 1(#!+(>(,0,*+1(2,*++?

@38$'8#:<(A:7703:85(,B0 (%')*?1(CDDE+

!!"

!!#

$

%&&'+'())

%&&'+'())

++,,++,,

++,,++,,

))

))

!"!

!""

###$$+%

$

###&&+%

&

,*

,*

-."'"(((

,0,$+1(2,$++(3$4563(78'(2"<$8'(7:"96=

* :3(4(<853$45$=(,#" 1(#!+(>(,0,*+1(2,*++?

#':3($."(8':;:549(:%4;"($8(&"(6:7703"6=(

;,?+(:3($."(45;9"(&"$F""5($F8(2"<$8'3

G5:38$'8#:<(A:7703:85(,H0(I(J4K4K(@LMNO*)+

Page 19: Geometric (Bio-) Modeling and Visualization

2010!"#$"%&'(%&!()*+$,$-(#,.&/-0+,.-1,$-(#

2#0$-$+$"&('&!()*+$,$-(#,.&,#3&4#5-#""%-#5&67-"#7"0

8"*,%$)"#$&('&!()*+$"%&67-"#7"0&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&9#-:"%0-$;&('&<"=,0&,$&>+0$-# !"#$"%&"'(()**7

!"#$%&'&$(!"#

!"#$%#&'())*'++,"'#- .-'"#$%#&'()*'++,"'#-

Page 20: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

• Surface: S: z = f(x,y) – explicit form• F(x,y,z) = 0 – implicit form

Principal Curvatures - I

•Tangent Space Basis:

•Normal:

•1st Fundamental Form: Measures the squared length of a tangent vector u by

Page 21: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Principal Curvatures - II•2nd Fundamental Form: Measures the change of unit normal along a unit tangent vector u via

•Normal change along a direction u: Curvature of the curve on the surface passing through the point p and having tangent u

•Principal Curvatures: Maximum and Minimum Normal Curvatures with the corresponding directions as principal curvature directions.

Page 22: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

•Principal Curvatures are the directions in the tangent space that optimize the quantity

Principal Curvatures - III

•They are the Eigenvalues/Eigenvectors of the matrix i.e. The Shape Operator S =

•S is also known as Weingarten Matrix W corresponding to the Weingarten Map

Page 23: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Level Set of Implicit Fn. (3D)

• Gradient:

• Normal:

• Level Set:

• Hessian:

Page 24: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Curvature

• Jacobian of Normal:

• Nonzero Eigenvalue/vectors of C gives principal curvatures.

• Surface M in 3D is the level set F(x,y,z) = 0 of the 3D Map F. Shape Operator S is given by

• Eigenvalues of S are c1, c2, 0 where c1, c2 are principal curvatures.

Page 25: Geometric (Bio-) Modeling and Visualization

Center for Computational VisualizationInstitute of Computational and Engineering SciencesDepartment of Computer Sciences University of Texas at Austin November 2007

Summary of Curvature Computations for • Surface M in 3D is the level set F(x,y,z) = 0 of the 3D Map F. The Shape Operator is extended by adding the extra constraint S(n) = 0 and thus S: R3 -> R3 is given by

• Surface M in 3D is explicitly given by z = f(x,y), equivalently by the level set F(x,y,z) = f(x,y) – z = 0.• The Shape Operator S: R2 -> R2 is given by

• Basis of Tangent Space

Page 26: Geometric (Bio-) Modeling and Visualization

Additional Reading • The references given below include the ones cited in

the lecture slides. Please check for pdf’s of these additional references on university computers from http://cvcweb.ices.utexas.edu/cvc/papers/papers.php

• C.Bajaj Tutorial Notes on “Multiscale, Bio-Modeling and Visualization”, Chap 2, 2010.