34
Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Embed Size (px)

Citation preview

Page 1: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Overview ofPropagating Interfaces

Donald Tanguay

October 30, 2002

Page 2: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Outline

• Propagating Interfaces

• Example: motion under curvature

• Functional Formulation

• Parametric Formulation

• Level Set Formulation

• Relationship to Snakes

Page 3: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Interfaces

An interface (or front) is a boundary between two regions: “inside” and “outside.”

In 2-D, an interface is a simple closed curve:

Page 4: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Propagating InterfacesHow does an interface evolve over time?At a specific moment, the speed function F (L, G, I) describesthe motion of the interface in the normal direction.

Page 5: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Propagating Interfaces

• Local – depend on local geometric information (e.g., curvature and normal direction)• Global – depend on the shape and position of the front (e.g., integrals along the front, heat diffusion)• Independent – do not depend on the shape of the front (e.g., an underlying fluid velocity that passively transports the front)

Speed F(L,G,I) is a function of 3 types of properties:

Page 6: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Motion Under Curvature

Example: Motion by curvature. Each piece moves perpendicularto the curve with speed proportional to the local curvature.

largepositivemotion

smallnegativemotion

Page 7: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Motion Under Curvature

Curvature κ is the inverse of theradius r of the osculating circle.

Page 8: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Motion Under Curvature

http://math.berkeley.edu/~sethian/Applets/java_curve_flow.html

Page 9: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Functional Representation

Eulerian framework: define fixed coordinate system on the world.For every world point x, there is (at most) one value y = ft(x).Falling snow example:

)(0 xfy t )(1 xfy t

Page 10: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Functional Representation

However, many simple shapes are multivalued;they are not functions regardless of the orientation of thecoordinate system.

Page 11: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Parametric RepresentationSpatially parameterize the curve x by s so that at time t the curve is xt(s), where 0 s S and the curve is closed: xt(0) = xt(S).

Points on initial curve.

Gradient (wrt time) is thespeed in normal direction.

Normal is perpendicularto curve, as is curvature.

Page 12: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Parametric RepresentationFor motion under curvature, speed F depends only onlocal curvature κ – the equation of motion is thus:

where curvature is

and the normal is

2/322 )()(

ss

sssssst yx

yxxys

2/1222/122 )()(

)(ss

Tss

ss

Tss

t yx

xy

yx

yx

x

xsn

),())(()( snsFsxt ttt

Page 13: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Particle MethodsIn order to compute, discretize the parameterization into moving particles which reconstruct the front.Known under a variety of names: marker particle techniques, string methods, nodal methods.

= # mesh particles

s

t

sS

= time step

= parameterization step

),( ni

ni yx = location of point iΔs at time nΔt

Δs

Page 14: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Particle MethodsTo construct a numerical algorithm, the derivatives areapproximated as central differences based on the Taylor series:

Page 15: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Particle Methods

Page 16: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Numerical Instability

Because Δs has dropped from Eq. 4.4, as neighboringparticles move closer together the quotient approaches 0/0,which is numerically unstable!

Uncontrollable oscillations stem from a feedback loop:1. Small errors in particle positions produce2. Local variations in approximated derivatives leading to3. Variations in computed particle velocities causing4. Uneven advancement of particles, yielding5. Larger errors in particle positions.

Page 17: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Numerical Instability

Decreasing time steps overcomes the instability but at additionalcomputational cost.

Varying the time step affects the final solution to an example problem:

BAD BETTER BEST

Page 18: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Numerical Instability

Typical, unappealing remedies that alter the motion equations in non-obvious ways:

• “Smooth” the speed function to keep particles apart

• Redistribute particles periodically

• Introduce a filter to remove the oscillations in particle positions

Page 19: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Changing Topology

Example: two fires merge into a single fire.

Page 20: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Changing Topology

Buoys!

In particle methods:• Difficult (and expensive) to detect and change the particle chains• Much more difficult as dimensionality increases

Page 21: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Difficulties With Particle Methods

• Instability

• Local singularities

• Management of particles: remove, redistribute, connect

Page 22: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Level Set FormulationRecast problem with one additional dimension – the distance from the interface.

),(0 yxz t

Page 23: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Changing Topology

),(1 yxz t

Page 24: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Level Set FormulationThe interface always lies at the zeroth level set of the function ,i.e., the interface is defined by the implicit equation t (x, y) = 0.

Page 25: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Initial Value FormulationDefine F as speed in normal direction:

)()()( txFtxntxF

A particle on the front with path x(t) is on the zero level set:

0)),(( ttx

0)( tx txdt

d

0)()),(( ttxttx

(1)

by chain rule:

(2)

(1) into (2): ),0,(given ,0 txF t

which is the continuous level set equation.

Page 26: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Simple Computational Scheme

01

nijij

nij

nij F

t

ionapproximat derivative spatial nijij

),,( tnjhihnij

h = spacing of uniform mesh

t = time step

(i, j) = grid nodes

• Discrete grid in x-y domain• Finite difference approximations to derivatives

The discrete level set equation:

Page 27: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Behavior at SingularitiesNaive formulation has bad behavior at “corner” singularities.

BAD GOOD

Page 28: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Viscosity SolutionsThe straightforward speed function (e.g., F = 1) causes swallowtail.Add a little curvature term (viscosity), and the curve is

mathemagically well-behaved!

)1(lim0

F-lysmall

Page 29: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Level Set Benefits

• Straightforward in higher dimensions

• Topological changes are natural

• Accurate computational schemes exist

• Intrinsic geometric properties are easy to determine (e.g., normal, curvature)

• Adaptive computational strategies improve efficiency

Page 30: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

• Functional explicit:– Inadequate model for most problems

• Parametric:– Adequate model, but

– Numerical instabilities

– Explicit topology makes changes difficult

• Level Set, functional implicit:– Adequate model, and

– Naturally handles changing topology, but

– Slow; needs efficiency improvements

Summary

Tt tsytsxsx ),(),()(

)(xfy t

0),( yxt

Page 31: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Summary

• Level sets is a general framework for accurately advancing an interface.

• Much of the challenge in particular interface problems is producing an adequate model of F.

Page 32: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Relation to Snakes

Snakes (active contours) have evolved greatly from the initial1988 paper of Kass, Witkin, and Terzopoulos.The term “snake” is now overloaded, and apparently thedifference from level sets is often exaggerated.

Two characteristics of snake formulations:• Neighbors connected into a chain• Additional “rod” or “spring” term in energy function attempts to keep neighbors apart to prevent degenerate chains

Page 33: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

Further Topics

• Level Set implementation

• Narrow Band Technique – optimization for the initial value formulation

• Fast Marching Methods – optimization for boundary value formulation, F > 0

• Art of Designing F

• Many example applications

Page 34: Overview of Propagating Interfaces Donald Tanguay October 30, 2002

References

M. Kass, A. Witkin, and D. Terzopoulos, “Snakes: ActiveContour Models,” IJCV, Jan. 1988, pp. 321-331.

J. A. Sethian. Level Set Methods and Fast Marching Methods.Cambridge University Press, 1999.

J. A. Sethian, “Tracking Interfaces with Level Sets,”American Scientist, May-June 1997.