Lecture 12 World Income Inequality: past, present and future.
Lecture Present
-
Upload
sivaloges-singam -
Category
Documents
-
view
219 -
download
0
Transcript of Lecture Present
-
8/2/2019 Lecture Present
1/27
Bayesian Modeling of Uncertaintyin Low-Level Vision
Paper by Richard Szeliski
International Journal of Computer Vision
(1990)
Presentation by Michael Ross
-
8/2/2019 Lecture Present
2/27
Bayesian methods for intrinsicimages
Dense fields and inverse problems (depthmaps, orientation, stereo matching).
Traditionally solved by energy minimization
or regularization methods.
Replace these methods with Bayesianmethods to get explicit modeling ofuncertainty, better understanding of models.
-
8/2/2019 Lecture Present
3/27
Intrinsic images
Finding denseinformation fieldsfrom sparse or
uncertain data.
Interpolation andaccounting for
uncertainty inmeasurement.
-
8/2/2019 Lecture Present
4/27
Energy methods
E u 1 E d u , d E p u
Ed u , d 12 i
c i u x i, y i d i
2
E p u
1
2u x
2u y
2dx dy
-
8/2/2019 Lecture Present
5/27
Discrete energy methods
E u 1
2
uTA u u
Tb c
Ed u , d
1
2 u
d
T
A d u
d
E p u
1
2
uTA p u
E u 1
2u u '
TA u u ' k
-
8/2/2019 Lecture Present
6/27
Energy implies probability
Squared error terms imply Gaussian noisemodels.
Smoothness (and discontinuity) can be
modeled as a Markov random field.
Using probability directly allows us access toinformation hidden by energy models(uncertainty modeling, model accuracy).
-
8/2/2019 Lecture Present
7/27
Markov random fields
p u i
u p u i
u j j
N i
A Markov random chain intwo dimensions. Mostcommonly used to calculatea MAP estimate, the
assignment whichmaximizes p(u|d).
-
8/2/2019 Lecture Present
8/27
Prior models
Describes p(u), our model of the world givenno data. In principle, we can calculate it fromthe MRF terms.
Gibbs distributions and sampling make thesecalculations simple and tractable. (Geman &Geman)
p u 1
Z pexp E p u
T p E p u E c u
-
8/2/2019 Lecture Present
9/27
Graph cliques
-
8/2/2019 Lecture Present
10/27
Sampling from prior models
p u i
u 1
Z pexp E p u i
u
T p
Thin plate on data. Sample thin plate.
-
8/2/2019 Lecture Present
11/27
Coarse to fine sampling.
Thin plate on data. Coarse-fine thin plate sample.
Why is this so much better?
-
8/2/2019 Lecture Present
12/27
Coarse to fine sampling.
Fourier analysis reveals that the surfacesdefined by thin-plate models are inherentlyfractal (self-similar at different resolutions).
Gibbs sampling at a single resolution is not apractical way to propagate the low-frequencycomponents of the random field.
Coarse to fine sampling is faster and producesmore representative samples.
-
8/2/2019 Lecture Present
13/27
Sensor models
Directly incorporated and easy to swap.
p d
u 1
Zdexp Ed u , d E d u , d
E di
u i, d i
Typical Gaussian sensor model (assuminguncorrelated errors).
p d i
u 1
2
i
exp
u i
d i 2
2
i
-
8/2/2019 Lecture Present
14/27
Contaminated gaussians
p d i
u 1
2
1
exp
u i
d i2
2
1
2
2
2
exp
u i
d i2
2
2
2
-
8/2/2019 Lecture Present
15/27
Virtual sensors
Use a lower-level vision algorithm (opticalflow, for example) as a dense data source.
Anandan and Weiss found that optical flow
errors depend on the type of local dataavailable (none, line, or corner).
This can be used to construct a higher-level
model that produces better flow estimates byexplicitly modeling these dependencies in thesensor distribution.
-
8/2/2019 Lecture Present
16/27
Posterior models
The output (intrinsic image) given the data -used to compute the MAP estimate of p(u | d).
Described by a Gibbs distribution (MRF) with
energy:E u E p u
E d u , d
Priormodel
Sensormodel
-
8/2/2019 Lecture Present
17/27
Loss functions
L
L u , u ' p u
d du
L
u , u ' p u
d du
Generic:
Map:
Loss functions allow for top-down influence.
-
8/2/2019 Lecture Present
18/27
Uncertainty estimatation
We can compute the covariance of theposterior models.
In the Gaussian sensor model case:
E u
1
2u u '
TA u u '
k
cov u
A
1
-
8/2/2019 Lecture Present
19/27
Stochastic variance estimation
Run the Gibbs sampler and estimate thevariance with a Monte Carlo algorithm.
Time averages are ergodic only over long
time frames, unless we use multiresolutionsampling.
1000 iterations 100 iterations
-
8/2/2019 Lecture Present
20/27
Dynamic models
Kalman filtering is a natural part of the MRFframework.
u
N u 0,P 0
d k H ku k
rk
u k Fku k 1
q k
Prior model
Sensor model
System model
-
8/2/2019 Lecture Present
21/27
Dynamic models
This work applies Kalman filtering to denseinformation, by using the sparse informationmatrices rather than the dense covariancematrices.
Ak
'
'
Ak
' '
Hk
TR
k
1H
k
T
b k'
'
b k' '
H kT
R k
1d k
uk
Ak
1b
k
-
8/2/2019 Lecture Present
22/27
Dynamic models
Why are the information matrices sparse?
In the prior model, we have local smoothness,so any column of the information matrix only
has non-zero values for the locallyneighboring points.
In the sensor model, we usually assume that
the sensor errors are uncorrelated, so thesensor information matrix is diagonal.
-
8/2/2019 Lecture Present
23/27
Applications
Incremental depth from motion, using theKalman filtering approach.
Motion estimation without correspondence
(discover the estimate that maximizes thelikelihood that two sets of points are drawnfrom the same smooth surface).
Maximum likelihood estimation for theregularization parameter.
-
8/2/2019 Lecture Present
24/27
Incremental depth from motion
Compute a displacement estimate usingcorrelation between frames.
Transform to disparity map using camera
motion, integrate with predicted map (Kalmanfiltering).
Regularize map.
Predict next map using known cameramotion.
-
8/2/2019 Lecture Present
25/27
Motion without correspondence
Surface through set #1 Surface through set #2 Surface through both
Minimize the distance between point set #2 and the surface through set#1 and the surface through both sets.
-
8/2/2019 Lecture Present
26/27
ML parameter estimation
p u
d exp
E d u , d E p u
p
2
p d 2
H P0H
T R
1
2exp
1
2d
TH P
0H
T R
1d
P 0 1
p
2
A p
Take the log and maximize...
-
8/2/2019 Lecture Present
27/27
Conclusion
Probability models encompass energy models.
We can draw samples to check their validity.
We can model sensor behavior easily.
Flexibility in choosing loss functions.
Direct calculation of error.