Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Total Variation and Related Methods II.

Slides:



Advertisements
Similar presentations
Is it rational or irrational?
Advertisements

Dates for term tests Friday, February 07 Friday, March 07
Duality for linear programming. Illustration of the notion Consider an enterprise producing r items: f k = demand for the item k =1,…, r using s components:
General Linear Model With correlated error terms  =  2 V ≠  2 I.
Longest Common Subsequence
Various Regularization Methods in Computer Vision Min-Gyu Park Computer Vision Lab. School of Information and Communications GIST.
Bregman Iterative Algorithms for L1 Minimization with
Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Total Variation and related Methods Segmentation.
Introduction to Variational Methods and Applications
Martin Burger Total Variation 1 Cetraro, September 2008 Variational Methods and their Analysis Questions: - Existence - Uniqueness - Optimality conditions.
1 12. Principles of Parameter Estimation The purpose of this lecture is to illustrate the usefulness of the various concepts introduced and studied in.
T HE POWER OF C ONVEX R ELAXATION : N EAR - OPTIMAL MATRIX COMPLETION E MMANUEL J. C ANDES AND T ERENCE T AO M ARCH, 2009 Presenter: Shujie Hou February,
Network Systems Lab. Korea Advanced Institute of Science and Technology No.1 Some useful Contraction Mappings  Results for a particular choice of norms.
1 Theory of Differentiation in Statistics Mohammed Nasser Department of Statistics.
Chaper 3 Weak Topologies. Reflexive Space.Separabe Space. Uniform Convex Spaces.
The Most Important Concept in Optimization (minimization)  A point is said to be an optimal solution of a unconstrained minimization if there exists no.
The Simple Linear Regression Model: Specification and Estimation
Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Computing Transmembrane Potentials from.
Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Total Variation and related Methods: Error.
Economics D10-1: Lecture 4 Classical Demand Theory: Preference-based approach to consumer behavior (MWG 3)
Lecture 1 Linear Variational Problems (Part I). 1. Motivation For those participants wondering why we start a course dedicated to nonlinear problems by.
Chebyshev Estimator Presented by: Orr Srour. References Yonina Eldar, Amir Beck and Marc Teboulle, "A Minimax Chebyshev Estimator for Bounded Error Estimation"
Minimaxity & Admissibility Presenting: Slava Chernoi Lehman and Casella, chapter 5 sections 1-2,7.
Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Total Variation and related Methods Numerical.
1 Introduction to Kernels Max Welling October (chapters 1,2,3,4)
Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Total Variation and Related Methods.
Error Estimation in TV Imaging Martin Burger Institute for Computational and Applied Mathematics European Institute for Molecular Imaging (EIMI) Center.
Background vs. foreground segmentation of video sequences = +
7(2) THE DUAL THEOREMS Primal ProblemDual Problem b is not assumed to be non-negative.
Optical Flow Estimation using Variational Techniques Darya Frolova.
Martin Burger Total Variation 1 Cetraro, September 2008 Numerical Schemes Wrap up approximate formulations of subgradient relation.
Definition and Properties of the Cost Function
1 10. Joint Moments and Joint Characteristic Functions Following section 6, in this section we shall introduce various parameters to compactly represent.
KKT Practice and Second Order Conditions from Nash and Sofer
Asaf Cohen (joint work with Rami Atar) Department of Mathematics University of Michigan Financial Mathematics Seminar University of Michigan March 11,
Example We can also evaluate a definite integral by interpretation of definite integral. Ex. Find by interpretation of definite integral. Sol. By the interpretation.
Linear Programming System of Linear Inequalities  The solution set of LP is described by Ax  b. Gauss showed how to solve a system of linear.
Extremal Problems of Information Combining Alexei Ashikhmin  Information Combining: formulation of the problem  Mutual Information Function for the Single.
MA4266 Topology Wayne Lawton Department of Mathematics S ,
Predicting Output from Computer Experiments Design and Analysis of Computer Experiments Chapter 3 Kevin Leyton-Brown.
I.3 Introduction to the theory of convex conjugated function.
PROBABILITY AND STATISTICS FOR ENGINEERING Hossein Sameti Department of Computer Engineering Sharif University of Technology Principles of Parameter Estimation.
Raeda Naamnieh 1. Outline Subdivision of Bezier Curves Restricted proof for Bezier Subdivision Convergence of Refinement Strategies 2.
Section 1.2 Gaussian Elimination. REDUCED ROW-ECHELON FORM 1.If a row does not consist of all zeros, the first nonzero number must be a 1 (called a leading.
Stein Unbiased Risk Estimator Michael Elad. The Objective We have a denoising algorithm of some sort, and we want to set its parameters so as to extract.
Probability Spaces A probability space is a triple (closed under Sample Space (any nonempty set), Set of Events a sigma-algebra over complementation and.
Joint Moments and Joint Characteristic Functions.
Linear & Nonlinear Programming -- Basic Properties of Solutions and Algorithms.
CDC Relative Entropy Applied to Optimal Control of Stochastic Uncertain Systems on Hilbert Space Nasir U. Ahmed School of Information Technology.
1 Kernel-class Jan Recap: Feature Spaces non-linear mapping to F 1. high-D space 2. infinite-D countable space : 3. function space (Hilbert.
Summary of the Last Lecture This is our second lecture. In our first lecture, we discussed The vector spaces briefly and proved some basic inequalities.
2.5 The Fundamental Theorem of Game Theory For any 2-person zero-sum game there exists a pair (x*,y*) in S  T such that min {x*V. j : j=1,...,n} =
Amir Yavariabdi Introduction to the Calculus of Variations and Optical Flow.
Theorem of Banach stainhaus and of Closed Graph
12. Principles of Parameter Estimation
Chapter 10 Optimal Control Homework 10 Consider again the control system as given before, described by Assuming the linear control law Determine the constants.
Proving that a Valid Inequality is Facet-defining
Quantum Two.
Chapter 2 Minimum Variance Unbiased estimation
Root-Locus Analysis (1)
Partly Verifiable Signals (c.n.)
§1-3 Solution of a Dynamical Equation
I.4 Polyhedral Theory (NW)
Charles University Charles University STAKAN III
I.4 Polyhedral Theory.
Proving that a Valid Inequality is Facet-defining
CIS 700: “algorithms for Big Data”
12. Principles of Parameter Estimation
Calculus In Infinite dimensional spaces
Numerical Methods for solutions of equations
Presentation transcript:

Martin Burger Institut für Numerische und Angewandte Mathematik European Institute for Molecular Imaging CeNoS Total Variation and Related Methods II

Martin Burger Total Variation 2 Cetraro, September 2008 Variational Methods and their Analysis We investigate the analysis of variational methods in imaging Most general form:

Martin Burger Total Variation 3 Cetraro, September 2008 Variational Methods and their Analysis Questions: - Existence - Uniqueness - Optimality conditions for solutions (-> numerical methods) - Structural properties of solutions - Asymptotic behaviour with respect to

Martin Burger Total Variation 4 Cetraro, September 2008 Variational Methods and their Analysis Two simplifying assumptions: -Noise is Gaussian (variance can be incorporated into ) - A is linear ´ Y Hilbert space

Martin Burger Total Variation 5 Cetraro, September 2008 TV Regularization Under the above assumptions we have

Martin Burger Total Variation 6 Cetraro, September 2008 Mean Value Technical simplification by eliminating mean value

Martin Burger Total Variation 7 Cetraro, September 2008 Mean Value Eliminate mean value Hence, minimum is attained among those functions with mean value c

Martin Burger Total Variation 8 Cetraro, September 2008 Mean Value We can minimize a-priori over the mean value and restrict the image to mean value zero W.r.o.g.

Martin Burger Total Variation 9 Cetraro, September 2008 Structure of BV 0 Equivalent norm

Martin Burger Total Variation 10 Cetraro, September 2008 Poincare-Inequality Proof. Assume does not hold. Then for each natural number n there is such that

Martin Burger Total Variation 11 Cetraro, September 2008 Poincare-Inequality Proof (ctd).

Martin Burger Total Variation 12 Cetraro, September 2008 Poincare-Inequality Proof (ctd).

Martin Burger Total Variation 13 Cetraro, September 2008 Dual Space Property Define

Martin Burger Total Variation 14 Cetraro, September 2008 Dual Space Property

Martin Burger Total Variation 15 Cetraro, September 2008 Dual Space Property

Martin Burger Total Variation 16 Cetraro, September 2008 Dual Space Property

Martin Burger Total Variation 17 Cetraro, September 2008 Dual Space Property

Martin Burger Total Variation 18 Cetraro, September 2008 Existence Basic ingredients of an existence proof are -Sequential lower semicontinuity - Compactness

Martin Burger Total Variation 19 Cetraro, September 2008 Existence What is the correct topology ?

Martin Burger Total Variation 20 Cetraro, September 2008 Lower Semicontinuity Compactness follows in the weak* topology. Lower semicontinuity ?

Martin Burger Total Variation 21 Cetraro, September 2008 Lower Semicontinuity

Martin Burger Total Variation 22 Cetraro, September 2008 Lower Semicontinuity

Martin Burger Total Variation 23 Cetraro, September 2008 Lower Semicontinuity First term: analogous proof implies

Martin Burger Total Variation 24 Cetraro, September 2008 Existence Theorem: Let J be sequentially lower semicontinuous and be compact. Then there exists a minimum of J Proof.

Martin Burger Total Variation 25 Cetraro, September 2008 Existence Proof (ctd). Due to compactness, there exists a subsequence, again denoted by such that By lower semicontinuity Hence, u is a minimizer

Martin Burger Total Variation 26 Cetraro, September 2008 Uniqueness Since the total variation is not strictly convex and definitely will not enforce uniqueness, the data term should do Proof: