Tomographic Approach for Sampling Multidimensional Signals with Finite Rate of Innovation Pancham Shukla and Pier Luigi Dragotti Communications and Signal.

Slides:



Advertisements
Similar presentations
Notes 6.6 Fundamental Theorem of Algebra
Advertisements

Shapelets Correlated with Surface Normals Produce Surfaces Peter Kovesi School of Computer Science & Software Engineering The University of Western Australia.
Compressive Sensing IT530, Lecture Notes.
Instructor: Yonina Eldar Teaching Assistant: Tomer Michaeli Spring 2009 Modern Sampling Methods
Contents 1. Introduction 2. UWB Signal processing 3. Compressed Sensing Theory 3.1 Sparse representation of signals 3.2 AIC (analog to information converter)
1. INTRODUCTION AND MOTIVATION Sampling is a fundamental step in obtaining sparse representation of signals (e.g. images, video) for applications such.
Image Denoising using Locally Learned Dictionaries Priyam Chatterjee Peyman Milanfar Dept. of Electrical Engineering University of California, Santa Cruz.
Compressed sensing Carlos Becker, Guillaume Lemaître & Peter Rennert
Data mining and statistical learning - lecture 6
Lecture05 Transform Coding.
Modern Sampling Methods Summary of Subspace Priors Spring, 2009.
Paul Heckbert Computer Science Department Carnegie Mellon University
Ch 6.2: Solution of Initial Value Problems
Basic Concepts and Definitions Vector and Function Space. A finite or an infinite dimensional linear vector/function space described with set of non-unique.
APPROXIMATING PDFs FOR MULTIPHASE PARTICULATE FLOW USING THE MAXIMUM ENTROPY METHOD Stephen J. Scott, John S. Shrimpton Thermofluids Section, Department.
CS Subdivision I: The Univariate Setting Peter Schröder.
Chapter 6 Numerical Interpolation
The Nyquist–Shannon Sampling Theorem. Impulse Train  Impulse Train (also known as "Dirac comb") is an infinite series of delta functions with a period.
DIGITAL SIGNAL PROCESSING IN ANALYSIS OF BIOMEDICAL IMAGES Prof. Aleš Procházka Institute of Chemical Technology in Prague Department of Computing and.
LIAL HORNSBY SCHNEIDER
Basic Principles of Computed Tomography Dr. Kazuhiko HAMAMOTO Dept. of Infor. Media Tech. Tokai University.
ENG4BF3 Medical Image Processing
Unitary Extension Principle: Ten Years After Zuowei Shen Department of Mathematics National University of Singapore.
Warm up Use the Rational Root Theorem to determine the Roots of : x³ – 5x² + 8x – 6 = 0.
1 1.1 © 2012 Pearson Education, Inc. Linear Equations in Linear Algebra SYSTEMS OF LINEAR EQUATIONS.
Numerical Methods Discrete Fourier Transform Part: Discrete Fourier Transform
Dr Martin Hendry, Dept of Physics and Astronomy University of Glasgow, UK Astronomical Data Analysis I 11 lectures, beginning autumn 2008.
SAMPLING SCHEMES FOR 2-D SIGNALS WITH FINITE RATE OF INNOVATION * Pancham Shukla Communications and Signal Processing Group Imperial College London  This.
UNIT - 4 ANALYSIS OF DISCRETE TIME SIGNALS. Sampling Frequency Harry Nyquist, working at Bell Labs developed what has become known as the Nyquist Sampling.
ORDINARY DIFFERENTIAL EQUATION (ODE) LAPLACE TRANSFORM.
1 Orthonormal Wavelets with Simple Closed-Form Expressions G. G. Walter and J. Zhang IEEE Trans. Signal Processing, vol. 46, No. 8, August 王隆仁.
Cs: compressed sensing
V. Space Curves Types of curves Explicit Implicit Parametric.
Digital Signal Processing
Extrapolation Models for Convergence Acceleration and Function ’ s Extension David Levin Tel-Aviv University MAIA Erice 2013.
EE369C Final Project: Accelerated Flip Angle Sequences Jan 9, 2012 Jason Su.
Shape Based Image Retrieval Using Fourier Descriptors Dengsheng Zhang and Guojun Lu Gippsland School of Computing and Information Technology Monash University.
Chapter #5 Pulse Modulation
Splines Vida Movahedi January 2007.
Boyce/DiPrima 9th ed, Ch 4.2: Homogeneous Equations with Constant Coefficients Elementary Differential Equations and Boundary Value Problems, 9th edition,
Signal Processing and Representation Theory Lecture 2.
Boyce/DiPrima 9 th ed, Ch 10.8: Laplace’s Equation Elementary Differential Equations and Boundary Value Problems, 9 th edition, by William E. Boyce and.
Estimation of Number of PARAFAC Components
Different types of wavelets & their properties Compact support Symmetry Number of vanishing moments Smoothness and regularity Denoising Using Wavelets.
Sampling and Interpolation
Zero of Polynomial Functions Factor Theorem Rational Zeros Theorem Number of Zeros Conjugate Zeros Theorem Finding Zeros of a Polynomial Function.
An Attempt to Extrapolate as a Research Fellow in Data Fusion Industrial multi-partner collaborative project - TATEM Pancham Shukla RA and PhD student.
Digital Image Processing
Topics 1 Specific topics to be covered are: Discrete-time signals Z-transforms Sampling and reconstruction Aliasing and anti-aliasing filters Sampled-data.
Instructor: Mircea Nicolescu Lecture 7
Ch 4.2: Homogeneous Equations with Constant Coefficients Consider the nth order linear homogeneous differential equation with constant, real coefficients:
Boyce/DiPrima 9 th ed, Ch 6.1: Definition of Laplace Transform Elementary Differential Equations and Boundary Value Problems, 9 th edition, by William.
Terahertz Imaging with Compressed Sensing and Phase Retrieval Wai Lam Chan Matthew Moravec Daniel Mittleman Richard Baraniuk Department of Electrical and.
1 Teaching Innovation - Entrepreneurial - Global The Centre for Technology enabled Teaching & Learning, N Y S S, India DTEL DTEL (Department for Technology.
Frequency domain analysis and Fourier Transform
Off-the-Grid Compressive Imaging: Recovery of Piecewise Constant Images from Few Fourier Samples Greg Ongie PhD Candidate Department of Applied Math and.
Audiovisual Communications Laboratory Sparse Sampling: Variations on a Theme by Shannon Martin Vetterli, EPFL & UC Berkeley Joint work with T.Blu (CUHK),
Sampling and Interpolation: Examples from the real world Martin Vetterli with Benjamin Bejar Haro, Mihailo Kolundzija and R.Parhizkar.
By Poornima Balakrishna Rajesh Ganesan George Mason University A Comparison of Classical Wavelet with Diffusion Wavelets.
0 Assignment 1 (Due: 10/2) Input/Output an image: (i) Design a program to input and output a color image. (ii) Transform the output color image C(R,G,B)
Linear systems: University of Illinois at Chicago ECE 427, Dr. D. Erricolo University of Illinois at Chicago ECE 427, Dr. D. Erricolo Consider two complex.
Super-resolution MRI Using Finite Rate of Innovation Curves Greg Ongie*, Mathews Jacob Computational Biomedical Imaging Group (CBIG) University of Iowa.
… Sampling … … Filtering … … Reconstruction …
Notes Over 3.4 The Rational Zero Test
Introduction to Diffraction Tomography
Basic Principles of Computed Tomography
Filtering and State Estimation: Basic Concepts
3.4 Zeros of Polynomial Functions: Real, Rational, and Complex
Signals and Systems Lecture 3
 = N  N matrix multiplication N = 3 matrix N = 3 matrix N = 3 matrix
Presentation transcript:

Tomographic Approach for Sampling Multidimensional Signals with Finite Rate of Innovation Pancham Shukla and Pier Luigi Dragotti Communications and Signal Processing Group Electrical and Electronic Engineering Department Imperial College, London SW7 2AZ, UK {p.shukla, This research was funded in part by EPSRC (UK). 1

1. Introduction Sampling is a fundamental step in obtaining sparse representation of signals (e.g. images, video) for applications such as coding, communication, and storage. Shannon’s classical sampling theory considers sampling of bandlimited signals using ‘sinc’ kernel. However, most real- world signals are nonbandlimited, and acquisition devices are non-ideal. Fortunately, recent research on Sampling signals with finite rate of innovation (FRI) [1,2] suggests the ways of sampling and perfect reconstruction of many 1-D nonbandlimited signals (e.g. Diracs, Piecewise polynomials) using a rich class of kernels (e.g. sinc, Gaussian, kernels that reproduce polynomials and exponentials, kernels with rational Fourier transform). The reconstruction is based on annihilating filter method (Prony’s method). 2

Contribution: We extend the results of FRI sampling [2] in higher dimensions using compactly supported kernels (e.g. B- splines) that reproduce polynomials (i.e., satisfy Strang-Fix conditions). Perfect reconstruction ? SignalSamplesKernel Earlier, we have shown that it possible to perfectly reconstruct many 2-D nonbandlimited signals (or shapes) from their samples [5]. In sequel to [5], here we show the sampling of more general FRI signals using the connection between Radon projections and moments [3]. 3

2. Sampling Framework The generic 2-D sampling setup (can be extended in n-D as well). Input signal Sampling Samples Acquisition device Sampling kernel 4

Sampling kernels and moments from samples We consider the kernels that satisfy Strang-Fix conditions, and therefore, reproduce polynomials up to certain degree That is, where and are the known coefficients. Scaling functions (from wavelet theory) and B-splines are examples of valid kernels. This makes it is possible to retrieve the continuous geometric moments of the original signal from its samples: 5

Polynomial of degree 1 along x Polynomial of degree 1 along y B-spline of degree 3 Reproduction of 2-D polynomials of degree 0 and 1 using B-spline kernel Polynomial of degree 0 6

3. Tomographic Approach Now we consider sampling of FRI signals such as 2-D polynomials with convex polygonal boundaries, and n-D Diracs and bilevel-convex polytopes using Radon transform and annihilating filter method. Radon transform projection of a 2-D function with compact support is given by: In fact, the Radon transform projections are obtained from the observed samples The reconstruction of the original signal is achieved by back-projection. 7

Annihilating Filter based Back-Projection (AFBP) algorithm Consider a case when is a 2-D polynomial of max. degree R-1 inside a convex polygonal closure with N corner points. In this case, we observe that 1. Each Radon projection is a 1-D piecewise polynomial of max. degree R and it can be decomposed into a stream of at most N differentiated Diracs using (R+1)- order derivatives. 2. Using Radon-moment connection of [3], we compute the moments of the differentiated Diracs from sample difference 8

3. Using these moments and the annihilating filter method [1], we retrieve Dirac locations and weights and therefore the projection itself [2]. 4. By iterating steps 1, 2 and 3 over N+1 distinct projection angles and then back-projecting the Dirac locations, we retrieve the convex polygonal closure [4]. 5.From the knowledge of the convex closure and max(R,N+1) Radon projections, we can determine the coefficients, and thus, the polynomial g(x,y) by solving a system of linear equations. Note that the sampling kernel must reproduce polynomials at least up to degree in this case. The AFBP algorithm can be extended for n-dimensional Diracs and bilevel-convex polytopes as well. 9

AFBP reconstruction of the 2-D polynomial with convex polygonal boundary. (a) The 2-D polynomial of degree R-1=0 inside convex polygon with N=5 corner points. (b) Radon transform projection Rg(t, theis a 1-D piecewise polynomial signal of degree R=1. (d) Second order derivative of the projection is a stream of N differentiated Diracs:]. 10

Simulation: Reconstruction of 2-D polynomial of degree R-1=0. Original signalSamplesReconst. of corner points Difference samples 11

4. References 1.M Vetterli, P Marziliano, and T Blu, ‘Sampling signals with finite rate of innovation,’ IEEE Trans. Sig. Proc., 50(6): , Jun P L Dragotti, M Vetterli, and T Blu, ‘Sampling moments and reconstructing signals of finite rate of innovation: Shannon meets Strang-Fix,’ IEEE Trans. Sig. Proc., Jun 2006, accepted. 3.P Milanfar, G Verghese, W Karl, and A Willsky, ‘Reconstructing polygons from moments with connections to array processing,’ IEEE Trans. Sig. Proc., 43(2): , Feb I Maravic and M Vetterli, ‘A sampling theorem for the Radon transform of finite complexity objects,’ Proc. IEEE ICASSP, , Orlando, Florida, USA, May P Shukla and P L Dragotti, ‘Sampling schemes for 2-D signals with finite rate of innovation using kernels that reproduce polynomials,’ Proc. IEEE ICIP, Genova, Italy, Sep