Temporal Scale-Spaces ScSp03

35
Temporal Scale Spaces Daniel Fagerström CVAP/NADA/KTH [email protected]

Transcript of Temporal Scale-Spaces ScSp03

Page 1: Temporal Scale-Spaces ScSp03

Temporal Scale Spaces

Daniel FagerströmCVAP/NADA/[email protected]

Page 2: Temporal Scale-Spaces ScSp03

Measurement of Temporal Signals

• Real time measurement• Temporal causality• Structure at different resolution• Well defined derivatives

Page 3: Temporal Scale-Spaces ScSp03

Earlier Approaches

• Koenderink (88), Florack (97), ter Haar Romeny et. al. (01) – Logaritmic mapping of the past time-line, then ordinary scale space

• Lindeberg & Fagerström (96) – Noncreation of local maxima with increasing scale

• Salden (99) – Diffusion on the temporal time-line with absorbing or reflecting boundary condition

Page 4: Temporal Scale-Spaces ScSp03

Current Approach

• Using axioms from Pauwels et. al. 95, replacing reflectional symmetry with temporal causality

Page 5: Temporal Scale-Spaces ScSp03

Space vs. Time

SpaceAll directions simultaneously available

TimeThe past can be memorized, the future can not

Memory

Page 6: Temporal Scale-Spaces ScSp03

Overview

• Motivation of axioms• Form of temporal scale space kernels• Time recursive realization• Numerical scheme• Comparison with earlier work

Page 7: Temporal Scale-Spaces ScSp03

Principles

• Non committed representation• Embodiment of symmetries in the

surroundings• Extended point

Page 8: Temporal Scale-Spaces ScSp03

Measurement

Posf : xf

Sensorf : f

Page 9: Temporal Scale-Spaces ScSp03

Meaurement

0|, 1 CL

•Choose well behaved sensor functions

Endf

fff ,

•Linearity

Page 10: Temporal Scale-Spaces ScSp03

Examples

dtttgf

•Regular distribution

0

•Dirac ”function”

fffff

•Derivation

Page 11: Temporal Scale-Spaces ScSp03

Causality

•Causality

0|1 tttt

•Measurement time

: t

Page 12: Temporal Scale-Spaces ScSp03

Embodiment of Geometry

• The structure of the sensor system should correspond to regularities in the environment

• Translation covariance: the measurement process is the same at each moment

• Scaling covariance: all time spans are treated in the same way

Page 13: Temporal Scale-Spaces ScSp03

Covariance

af

axfD

xafxfSaxfxfT

a

a

a

11 :Dilatation

:Scaling :nTranslatio

•Transformation groups

aa

aa

DffSTffT

•Corresponding action on a sensor

Page 14: Temporal Scale-Spaces ScSp03

Orbits

0','' tt TDT

Page 15: Temporal Scale-Spaces ScSp03

Convolution

'

'

f

dsstsfdstssfTf t

•Definition

ff

•Associativity

Page 16: Temporal Scale-Spaces ScSp03

Cascade Property (Semi-Group)

Page 17: Temporal Scale-Spaces ScSp03

Extended Point

•Unit area1

•Positivity 00 ff

ss 0

lim

Page 18: Temporal Scale-Spaces ScSp03

Temporal Scale Space Kernels

• Continuity• Positivity• Unit area• Causality• Dilatation covariance• Convolution semi-groupPauwels et al used the same axioms, with (4) replaced by reflexion

symmetry, to characterize spatial scale-spaces.

Page 19: Temporal Scale-Spaces ScSp03

Characterization

Laplace transform

ˆˆˆ

Semi-group and causality

:,ˆ ges sg

Continuity

(Cauchy’s functional equation)

Page 20: Temporal Scale-Spaces ScSp03

Characterization (Cont) sges

ˆ

Dilatation covariance

sesˆ

Positivity

(Bernstein)

10

Page 21: Temporal Scale-Spaces ScSp03

Temporal Scale Space Kernels0.3 0.5 0.8

set -1, L

Page 22: Temporal Scale-Spaces ScSp03

Extremal Stable Density Functions

•Limit densities for sum of stochastic variables (Levy)

•Infinite first and second moments

•Increasing popularity in fysics, finance …

Page 23: Temporal Scale-Spaces ScSp03

Explicit Form

0,

,1

23

2

,21

,0

!1

24exp4

k

kk

t

tkkt

t

tt

kt

tt

tt

Page 24: Temporal Scale-Spaces ScSp03

Markov Property?

• The memory is the only access to the past• Convolution directly with the input signal is

unrealistic• Need for an evaluation equation!

t

L(0,)=0

L(t,)=f(t)

LALt tf

tLftL

,,,,

Page 25: Temporal Scale-Spaces ScSp03

Fractional Derivatives

fxfsfxf xx ,

1, L

•Linear

•Equal to ordinary derivatives for integer order

xxx

•Generalized Leibniz property

gfgf kx

kx

kx

0 k

Page 26: Temporal Scale-Spaces ScSp03

Evaluation Equations

,

1

-1,

expLxpe

L

t

s

set

tftLL

LL t

,lim0,0

0

,•Signaling equation for

tftLL

LLt

,lim0,0

0

22,

•Fractional Brownian motion

•Diffusion for

L

LLL

t

tt

2,

,,

2

tftLL

LLt

,lim0,0

0

1,

•Scale space as state

•Diffusion for

Page 27: Temporal Scale-Spaces ScSp03

Uniquenes of the Signaling Equation

tftLL

LLt

,lim0,0

0

1,

•Locality: is a differential operator if 1/ is an integer

•Positivity: have a positive Greens function if 1/

•Locality and positivity is only satisfied for

L1

,

L1

,

Page 28: Temporal Scale-Spaces ScSp03

Signaling Equation

tftL

LLLt

,lim0,0

0

2

•The temporal scale space is the only memory of earlier input

•The memory diffuses over time

Page 29: Temporal Scale-Spaces ScSp03

Numerical Scheme

t

3/2-1/2

•Implicit scheme

•Second order stability in time and scale

•4(add)+4(multiply)+2(divide) per mesh point

Page 30: Temporal Scale-Spaces ScSp03

Example

Page 31: Temporal Scale-Spaces ScSp03

Example

Page 32: Temporal Scale-Spaces ScSp03

Comparison with Previous Work

• Florack (97) requires the measurement kernel to be a positive Schwartz test function, this rules out all stable density functions except Gaussians

• Ter Haar Romeny et. al. (01) requires the measurement kernel to have finite first and second moment, only fulfilled for Gaussians, extremal (causal) stable density functions even has infinite first moment

Page 33: Temporal Scale-Spaces ScSp03

Comparison with Earlier Work

• Koenderink (88) and Lindeberg & Fagerström (96) uses stronger requirements (than semi-group) on causality in the resolution domain

• Salden (99) does not require translational covariance on the kernel only on the generator

Page 34: Temporal Scale-Spaces ScSp03

Comparison with Earlier Work

• Lindeberg & Fagerström (96) and the current approach are the only ones that has a (known) time recursive formulation

• Lindeberg & Fagerström (96) requires either time or scale to be discrete and lack scale covariance

• The kernels in the current approach are less well localized than in the other approaches

Page 35: Temporal Scale-Spaces ScSp03

Conclusion

• Causal scale invariant convolution semi group

• Time recursive realization• Signaling equation• Efficient numerical scheme