The 2D orientation is unique through Principal Moments Analysis

Size: px
Start display at page:

Download "The 2D orientation is unique through Principal Moments Analysis"

Transcription

1 The 2D orientation is unique through Principal Moments Analysis João F. P. Crespo Pedro M. Q. Aguiar Institute for Systems and Robotics / IST Lisboa, Portugal September, 2010

2 Outline 1 Motivation 2 Moments for shape representation 3 Shape normalization through Principal Moments Analysis 4 Extension to gray-level images 5 Experiments 6 Conclusion

3 Motivation Representing 2D shapes for classification Shape diversity: For us, shapes are arbitrary sets of 2D points

4 Motivation Shape normalization Normalizing w.r.t. translation and scale is simple: Sampling density and unknown point labels also tratable: Problem: unknown shape orientation

5 Motivation PCA-based orientation Only the principal axis is determined, not its direction: Not a serious problem for elongated shapes. However...

6 Motivation Ambiguity in PCA-based orientation Many shapes do not have a stable principal axis For rotationally symmetric shapes, PCA can not be used at all: This the problem we address in the paper

7 Motivation Previous work Seeking a reasonable geometric orientation : mirror-symmetry axes [Atallah,1985][Marola,1989] universal principal axis [Lin, 1993] generalized principal axis [Tsai and Chou, 1991][Zunic et al, 2006] Detection of symmetry and fold number [Lin et al, 1994] [Shen and Ip, 1999] [Derrode and Ghorbel, 2004] [Prasad and Yegnanarayana, 2004] Requiring the exhaustive search for an angle maximizing a given measure (without any guarantee of uniqueness) [Ha and Moura, 2005] More theoretically sustained methods are based on moments: CMs [Abu-Mostafa and Psaltis, 1985][Teh and Chin,1988] GCMs [Shen and Ip, 1997][Shen and Ip, 1999] An insightful recent review: [Flusser, Suk, and Zitova, 2009] This work: overcoming limitations of previous works

8 Moments for shape representation Problem definition 2D shape given by an N-dimensional complex vector z = [ z 1 z 2 z N ] T, (zn = x n + jy n ) We seek an orientation angle θ(z) s.t. any shape z is brought to its normalized version j θ(z) w(z) = ze To guarantee the desired invariance, i.e., w ( ze jφ) = w(z), it suffices that ( φ, θ ze jφ) = θ(z) + φ (add prop the orientation angle of a rotated shape is equal to the sum of the orientation angle of the original shape with the rotation angle)

9 Moments for shape representation Shape moments We define and compute θ(z) using power sums: µ k (z) = z1 k + z2 k + + zn k, k {1, 2, 3,...} {µ 1, µ 2,..., µ N } describes univocally z [Kanatani, 1990], thus forms a complete set of invariants over the permutation group, or, equivalently, is maximally invariant to permutations A much smaller set suffices to discriminate in practice, having been called Principal Moments (PMs) [Crespo and Aguiar, 2009] We thus call our approach Principal Moments Analysis (PMA) Translation, scale, and sampling invariance is obtained by working with N (z z) / z z, rather than directly with z

10 Shape normalization through Principal Moments Analysis Arguments of moments of rotated shapes Shape rotation propagates to the moments in a nice way: µ k (ze jφ ) = µ k (z)e jkφ The choice θ(z) = arg µ 1 (z) satisfies the add prop θ(ze jφ ) = arg µ 1 (ze jφ ) = arg µ 1 (z) + φ = θ(z) + φ but it is useless in practice when µ 1 = 0 Equivalent to zeroing the argument of the first-order moment of the rotationally normalized shape Our approach is to generalize by doing the same to the k th : arg µ k (ze j θ(z)) = 0

11 Shape normalization through Principal Moments Analysis Zeroing arguments of moments Imposing arg µ 2 ( ze j θ(z) ) = 0 : arg n z 2 n e 2jθ(z) = 0 arg µ 2 (z) 2θ(z) = 0 θ(z) = arg µ 2(z) 2 Imposing arg µ k ( ze j θ(z) ) = 0 : θ(z) = arg µ 2(z) 2 + π (PCA) θ(z) = arg µ k(z) k + 2π k l, l {0, 1,..., k 1} Ambiguity: there are k distinct values of θ(z) that annihilate the argument of µ k (ze j θ(z) ) Pick a solution that satisfies the add prop

12 Shape normalization through Principal Moments Analysis Principal Moments Analysis Pick l s.t. θ(z) = arg µ k (z) + 2πl satisfies the add prop θ(ze jφ ) = θ(z) + φ; k k rotation propagation: µ k (ze jφ ) = µ k (z)e jkφ The argument of the k th moment of a rotated shape is arg µ k (ze jφ ) = arg µ k (z) + kφ + 2πˆl, where ˆl is s.t. arg µ k (ze jφ ) and arg µ k (z) are in ( π, π] The normalization angle of the rotated shape is then θ(ze jφ ) = arg µ k(z) k + φ + 2π k (l + ˆl) (1) If we choose any fixed l, the add prop would require ˆl = 0, which fails to guarantee that the arguments fall in ( π, π] Solution: select a value for l that depends on the shape, l(z)

13 Shape normalization through Principal Moments Analysis Principal Moments Analysis The normalization angle is θ(z) = arg µ k (z) k + 2πl(z) k with l(z) {0, 1,..., k 1} Consider a supplementary moment µ m, with k and m coprime; the normalization angle θ(z) uses arg µ k (z) and arg µ m (z) The values of arg µ m ( ze j θ(z) ), with l(z) {0, 1,..., k 1}}, are spaced by intervals of length 2π/k PMA: choose l(z) s.t. arg µ m ( ze j θ(z) ) falls within an arbitrary but fixed interval, e.g., [0, 2π/k) The choice for l(z) is unambiguous (uniqueness) The normalization angle θ(z) satisfies the add prop (rotation invariance) All shapes have more than one nonzero moment (universality)

14 Shape normalization through Principal Moments Analysis Rotational symmetry Are there always coprime (k, m) s. t. µ k 0 and µ m 0? No! It may occur γ = gcd{l : µ l 0} > 1 Considering PMs as coefficients of a Fourier series, it is simple to conclude that this is equivalent to a γ-fold rotational symmetry, i.e., that the shape is invariant to rotations of 2π/γ All normalization angles θ + ˆk 2π/γ lead to the same result and θ is computed by using PMA with decimated PMs

15 Extension to gray-level images Extension to gray-level images and improved robustness We generalize the power sums to the corresponding moments of continuous images: + µ k (g) = (x + jy) k g(x, y) dx dy, k {0, 1, 2,...} In what respects to representation, the generalization loses completeness: in opposition to the case of a set of points, the PMs {µ k } do not determine g(x, y) univocally Naturally, the lack of completeness does not impede the usage of PMA to normalize continuous images w.r.t. orientation To improve robustness, we integrate the contributions of several pairs {(k i, m i )} by computing the angle as the (angular) weighted average θ(z) = arg i p i e j θ i (z)

16 The 2D orientation is unique through Principal Moments Analysis Experiments Direction disambiguity and rotational symmetric examples

17 Experiments Normalization angle accuracy

18 The 2D orientation is unique through Principal Moments Analysis Experiments Robustness to sampling density

19 Experiments Normalization of grey-level images

20 Experiments Normalization of grey-level images

21 Experiments Normalization of grey-level images

22 Experiments Failure cases Grey-level images that appear to be rotationally symmetric Must be carefully constructed, e.g., the image f (r, θ) = R(r) (cos θ + cos 2θ), with a particular R(r), has moments µ 0 = 0, µ 1 = 0, µ 2 0, µ 3 = 0, µ 4 = 0, µ 5 = 0, µ 6 = 0,

23 Conclusion Summary The paper presents a new algorithm PMA to normalize 2D shapes (arbitrary sets of 2D points) w.r.t. orientation PMA computes an unambiguous orientation angle for any shape, including rotationally symmetric ones PMA uses a set of moments, or power sums, that fully describe the shape PMA can be applied to grey-level images but the completeness is lost Experiments illustrate robustness

Revisiting Complex Moments For 2D Shape Representation and Image Normalization

Revisiting Complex Moments For 2D Shape Representation and Image Normalization IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Revisiting Complex Moments For 2D Shape Representation and Image Normalization João B. F. P. Crespo and Pedro M. Q. Aguiar, Senior Member, IEEE Abstract When comparing

More information

Affine Normalization of Symmetric Objects

Affine Normalization of Symmetric Objects Affine Normalization of Symmetric Objects Tomáš Suk and Jan Flusser Institute of Information Theory and Automation, Academy of Sciences of the Czech Republic, Pod vodárenskou věží 4, 182 08 Prague 8, Czech

More information

Science Insights: An International Journal

Science Insights: An International Journal Available online at http://www.urpjournals.com Science Insights: An International Journal Universal Research Publications. All rights reserved ISSN 2277 3835 Original Article Object Recognition using Zernike

More information

A Constraint Relationship for Reflectional Symmetry and Rotational Symmetry

A Constraint Relationship for Reflectional Symmetry and Rotational Symmetry A Constraint Relationship for Reflectional Symmetry and Rotational Symmetry Dinggang Shen, Horace H. S. Ip 2 and Eam Khwang Teoh 3 Department of Radiology, Johns Hopkins University Email: dgshen@cbmv.jhu.edu

More information

Constraint relationship for reflectional symmetry and rotational symmetry

Constraint relationship for reflectional symmetry and rotational symmetry Journal of Electronic Imaging 10(4), 1 0 (October 2001). Constraint relationship for reflectional symmetry and rotational symmetry Dinggang Shen Johns Hopkins University Department of Radiology Baltimore,

More information

Target 6.1 The student will be able to use l Hôpital s Rule to evaluate indeterminate limits. lim. lim. 0, then

Target 6.1 The student will be able to use l Hôpital s Rule to evaluate indeterminate limits. lim. lim. 0, then Target 6.1 The student will be able to use l Hôpital s Rule to evaluate indeterminate limits. Recall from Section 2.1 Indeterminate form is when lim. xa g( Previously, we tried to reduce and then re-evaluate

More information

3-D Projective Moment Invariants

3-D Projective Moment Invariants Journal of Information & Computational Science 4: * (2007) 1 Available at http://www.joics.com 3-D Projective Moment Invariants Dong Xu a,b,c,d,, Hua Li a,b,c a Key Laboratory of Intelligent Information

More information

Definition 1. A set V is a vector space over the scalar field F {R, C} iff. there are two operations defined on V, called vector addition

Definition 1. A set V is a vector space over the scalar field F {R, C} iff. there are two operations defined on V, called vector addition 6 Vector Spaces with Inned Product Basis and Dimension Section Objective(s): Vector Spaces and Subspaces Linear (In)dependence Basis and Dimension Inner Product 6 Vector Spaces and Subspaces Definition

More information

Regional Solution of Constrained LQ Optimal Control

Regional Solution of Constrained LQ Optimal Control Regional Solution of Constrained LQ Optimal Control José DeDoná September 2004 Outline 1 Recap on the Solution for N = 2 2 Regional Explicit Solution Comparison with the Maximal Output Admissible Set 3

More information

Vector-attribute filters

Vector-attribute filters Vector-attribute filters Erik R. Urbach, Niek J. Boersma, and Michael H.F. Wilkinson Institute for Mathematics and Computing Science University of Groningen The Netherlands April 2005 Outline Purpose Binary

More information

Lecture course on crystallography, 2015 Lecture 5: Symmetry in crystallography

Lecture course on crystallography, 2015 Lecture 5: Symmetry in crystallography Dr Semën Gorfman Department of Physics, University of SIegen Lecture course on crystallography, 2015 Lecture 5: Symmetry in crystallography What is symmetry? Symmetry is a property of an object to stay

More information

= 10 such triples. If it is 5, there is = 1 such triple. Therefore, there are a total of = 46 such triples.

= 10 such triples. If it is 5, there is = 1 such triple. Therefore, there are a total of = 46 such triples. . Two externally tangent unit circles are constructed inside square ABCD, one tangent to AB and AD, the other to BC and CD. Compute the length of AB. Answer: + Solution: Observe that the diagonal of the

More information

Payment Rules for Combinatorial Auctions via Structural Support Vector Machines

Payment Rules for Combinatorial Auctions via Structural Support Vector Machines Payment Rules for Combinatorial Auctions via Structural Support Vector Machines Felix Fischer Harvard University joint work with Paul Dütting (EPFL), Petch Jirapinyo (Harvard), John Lai (Harvard), Ben

More information

WALLPAPER GROUPS. Julija Zavadlav

WALLPAPER GROUPS. Julija Zavadlav WALLPAPER GROUPS Julija Zavadlav Abstract In this paper we present the wallpaper groups or plane crystallographic groups. The name wallpaper groups refers to the symmetry group of periodic pattern in two

More information

Rotations in Quantum Mechanics

Rotations in Quantum Mechanics Rotations in Quantum Mechanics We have seen that physical transformations are represented in quantum mechanics by unitary operators acting on the Hilbert space. In this section, we ll think about the specific

More information

Lecture 4 Quantum mechanics in more than one-dimension

Lecture 4 Quantum mechanics in more than one-dimension Lecture 4 Quantum mechanics in more than one-dimension Background Previously, we have addressed quantum mechanics of 1d systems and explored bound and unbound (scattering) states. Although general concepts

More information

More On Carbon Monoxide

More On Carbon Monoxide More On Carbon Monoxide E = 0.25 ± 0.05 ev Electron beam results Jerry Gilfoyle The Configurations of CO 1 / 26 More On Carbon Monoxide E = 0.25 ± 0.05 ev Electron beam results Jerry Gilfoyle The Configurations

More information

Statistical Pattern Recognition

Statistical Pattern Recognition Statistical Pattern Recognition Feature Extraction Hamid R. Rabiee Jafar Muhammadi, Alireza Ghasemi, Payam Siyari Spring 2014 http://ce.sharif.edu/courses/92-93/2/ce725-2/ Agenda Dimensionality Reduction

More information

Optimal Fault-Tolerant Configurations of Thrusters

Optimal Fault-Tolerant Configurations of Thrusters Optimal Fault-Tolerant Configurations of Thrusters By Yasuhiro YOSHIMURA ) and Hirohisa KOJIMA, ) ) Aerospace Engineering, Tokyo Metropolitan University, Hino, Japan (Received June st, 7) Fault tolerance

More information

Lecture Note 1: Background

Lecture Note 1: Background ECE5463: Introduction to Robotics Lecture Note 1: Background Prof. Wei Zhang Department of Electrical and Computer Engineering Ohio State University Columbus, Ohio, USA Spring 2018 Lecture 1 (ECE5463 Sp18)

More information

A New Efficient Method for Producing Global Affine Invariants

A New Efficient Method for Producing Global Affine Invariants A New Efficient Method for Producing Global Affine Invariants Esa Rahtu, Mikko Salo 2, and Janne Heikkilä Machine Vision Group, Department of Electrical and Information Engineering, P.O. Box 45, 94 University

More information

Orientation Estimation in Ambiguous Neighbourhoods Mats T. Andersson & Hans Knutsson Computer Vision Laboratory Linköping University 581 83 Linköping, Sweden Abstract This paper describes a new algorithm

More information

FORMULATION OF THE LEARNING PROBLEM

FORMULATION OF THE LEARNING PROBLEM FORMULTION OF THE LERNING PROBLEM MIM RGINSKY Now that we have seen an informal statement of the learning problem, as well as acquired some technical tools in the form of concentration inequalities, we

More information

Linear Dimensionality Reduction

Linear Dimensionality Reduction Outline Hong Chang Institute of Computing Technology, Chinese Academy of Sciences Machine Learning Methods (Fall 2012) Outline Outline I 1 Introduction 2 Principal Component Analysis 3 Factor Analysis

More information

Final exam: CS 663, Digital Image Processing, 21 st November

Final exam: CS 663, Digital Image Processing, 21 st November Final exam: CS 663, Digital Image Processing, 21 st November Instructions: There are 180 minutes for this exam (5:30 pm to 8:30 pm). Answer all 8 questions. This exam is worth 25% of the final grade. Some

More information

PRINCIPAL COMPONENTS ANALYSIS

PRINCIPAL COMPONENTS ANALYSIS 121 CHAPTER 11 PRINCIPAL COMPONENTS ANALYSIS We now have the tools necessary to discuss one of the most important concepts in mathematical statistics: Principal Components Analysis (PCA). PCA involves

More information

Sparse Subspace Clustering

Sparse Subspace Clustering Sparse Subspace Clustering Based on Sparse Subspace Clustering: Algorithm, Theory, and Applications by Elhamifar and Vidal (2013) Alex Gutierrez CSCI 8314 March 2, 2017 Outline 1 Motivation and Background

More information

Quantum Mechanics II

Quantum Mechanics II Quantum Mechanics II Prof. Boris Altshuler April, 20 Lecture 2. Scattering Theory Reviewed Remember what we have covered so far for scattering theory. We separated the Hamiltonian similar to the way in

More information

Normed & Inner Product Vector Spaces

Normed & Inner Product Vector Spaces Normed & Inner Product Vector Spaces ECE 174 Introduction to Linear & Nonlinear Optimization Ken Kreutz-Delgado ECE Department, UC San Diego Ken Kreutz-Delgado (UC San Diego) ECE 174 Fall 2016 1 / 27 Normed

More information

Notes: Most of the material presented in this chapter is taken from Bunker and Jensen (2005), Chap. 3, and Atkins and Friedman, Chap. 5.

Notes: Most of the material presented in this chapter is taken from Bunker and Jensen (2005), Chap. 3, and Atkins and Friedman, Chap. 5. Chapter 5. Geometrical Symmetry Notes: Most of the material presented in this chapter is taken from Bunker and Jensen (005), Chap., and Atkins and Friedman, Chap. 5. 5.1 Symmetry Operations We have already

More information

Conditional Distributions

Conditional Distributions Conditional Distributions The goal is to provide a general definition of the conditional distribution of Y given X, when (X, Y ) are jointly distributed. Let F be a distribution function on R. Let G(,

More information

Nuclear models: Collective Nuclear Models (part 2)

Nuclear models: Collective Nuclear Models (part 2) Lecture 4 Nuclear models: Collective Nuclear Models (part 2) WS2012/13: Introduction to Nuclear and Particle Physics,, Part I 1 Reminder : cf. Lecture 3 Collective excitations of nuclei The single-particle

More information

A Selective Review of Sufficient Dimension Reduction

A Selective Review of Sufficient Dimension Reduction A Selective Review of Sufficient Dimension Reduction Lexin Li Department of Statistics North Carolina State University Lexin Li (NCSU) Sufficient Dimension Reduction 1 / 19 Outline 1 General Framework

More information

Chapter 12. Linear Molecules

Chapter 12. Linear Molecules Chapter 1. Linear Molecules Notes: Most of the material presented in this chapter is taken from Bunker and Jensen (1998), Chap. 17. 1.1 Rotational Degrees of Freedom For a linear molecule, it is customary

More information

CP1 REVISION LECTURE 3 INTRODUCTION TO CLASSICAL MECHANICS. Prof. N. Harnew University of Oxford TT 2017

CP1 REVISION LECTURE 3 INTRODUCTION TO CLASSICAL MECHANICS. Prof. N. Harnew University of Oxford TT 2017 CP1 REVISION LECTURE 3 INTRODUCTION TO CLASSICAL MECHANICS Prof. N. Harnew University of Oxford TT 2017 1 OUTLINE : CP1 REVISION LECTURE 3 : INTRODUCTION TO CLASSICAL MECHANICS 1. Angular velocity and

More information

Properties of Rational and Irrational Numbers

Properties of Rational and Irrational Numbers Properties of Rational and Irrational Numbers September 8, 2016 Definition: The natural numbers are the set of numbers N = {1, 2, 3,...}, and the integers are the set of numbers Z = {..., 2, 1, 0, 1, 2,...}.

More information

Overview of Statistical Tools. Statistical Inference. Bayesian Framework. Modeling. Very simple case. Things are usually more complicated

Overview of Statistical Tools. Statistical Inference. Bayesian Framework. Modeling. Very simple case. Things are usually more complicated Fall 3 Computer Vision Overview of Statistical Tools Statistical Inference Haibin Ling Observation inference Decision Prior knowledge http://www.dabi.temple.edu/~hbling/teaching/3f_5543/index.html Bayesian

More information

Chem Symmetry and Introduction to Group Theory. Symmetry is all around us and is a fundamental property of nature.

Chem Symmetry and Introduction to Group Theory. Symmetry is all around us and is a fundamental property of nature. Symmetry and Introduction to Group Theory Symmetry is all around us and is a fundamental property of nature. Symmetry and Introduction to Group Theory The term symmetry is derived from the Greek word symmetria

More information

Wavelet Analysis. Willy Hereman. Department of Mathematical and Computer Sciences Colorado School of Mines Golden, CO Sandia Laboratories

Wavelet Analysis. Willy Hereman. Department of Mathematical and Computer Sciences Colorado School of Mines Golden, CO Sandia Laboratories Wavelet Analysis Willy Hereman Department of Mathematical and Computer Sciences Colorado School of Mines Golden, CO 8040-887 Sandia Laboratories December 0, 998 Coordinate-Coordinate Formulations CC and

More information

Lecture 4 Quantum mechanics in more than one-dimension

Lecture 4 Quantum mechanics in more than one-dimension Lecture 4 Quantum mechanics in more than one-dimension Background Previously, we have addressed quantum mechanics of 1d systems and explored bound and unbound (scattering) states. Although general concepts

More information

18.03 LECTURE NOTES, SPRING 2014

18.03 LECTURE NOTES, SPRING 2014 18.03 LECTURE NOTES, SPRING 2014 BJORN POONEN 7. Complex numbers Complex numbers are expressions of the form x + yi, where x and y are real numbers, and i is a new symbol. Multiplication of complex numbers

More information

Optimization Problems

Optimization Problems Optimization Problems The goal in an optimization problem is to find the point at which the minimum (or maximum) of a real, scalar function f occurs and, usually, to find the value of the function at that

More information

The six vertex model is an example of a lattice model in statistical mechanics. The data are

The six vertex model is an example of a lattice model in statistical mechanics. The data are The six vertex model, R-matrices, and quantum groups Jethro van Ekeren. 1 The six vertex model The six vertex model is an example of a lattice model in statistical mechanics. The data are A finite rectangular

More information

Rotation of Axes. By: OpenStaxCollege

Rotation of Axes. By: OpenStaxCollege Rotation of Axes By: OpenStaxCollege As we have seen, conic sections are formed when a plane intersects two right circular cones aligned tip to tip and extending infinitely far in opposite directions,

More information

PCA & ICA. CE-717: Machine Learning Sharif University of Technology Spring Soleymani

PCA & ICA. CE-717: Machine Learning Sharif University of Technology Spring Soleymani PCA & ICA CE-717: Machine Learning Sharif University of Technology Spring 2015 Soleymani Dimensionality Reduction: Feature Selection vs. Feature Extraction Feature selection Select a subset of a given

More information

JUST THE MATHS UNIT NUMBER 6.2. COMPLEX NUMBERS 2 (The Argand Diagram) A.J.Hobson

JUST THE MATHS UNIT NUMBER 6.2. COMPLEX NUMBERS 2 (The Argand Diagram) A.J.Hobson JUST THE MATHS UNIT NUMBER 6.2 COMPLEX NUMBERS 2 (The Argand Diagram) by A.J.Hobson 6.2.1 Introduction 6.2.2 Graphical addition and subtraction 6.2.3 Multiplication by j 6.2.4 Modulus and argument 6.2.5

More information

8. Prime Factorization and Primary Decompositions

8. Prime Factorization and Primary Decompositions 70 Andreas Gathmann 8. Prime Factorization and Primary Decompositions 13 When it comes to actual computations, Euclidean domains (or more generally principal ideal domains) are probably the nicest rings

More information

Screw Theory and its Applications in Robotics

Screw Theory and its Applications in Robotics Screw Theory and its Applications in Robotics Marco Carricato Group of Robotics, Automation and Biomechanics University of Bologna Italy IFAC 2017 World Congress, Toulouse, France Table of Contents 1.

More information

1/30. Rigid Body Rotations. Dave Frank

1/30. Rigid Body Rotations. Dave Frank . 1/3 Rigid Body Rotations Dave Frank A Point Particle and Fundamental Quantities z 2/3 m v ω r y x Angular Velocity v = dr dt = ω r Kinetic Energy K = 1 2 mv2 Momentum p = mv Rigid Bodies We treat a rigid

More information

Pattern Recognition 2

Pattern Recognition 2 Pattern Recognition 2 KNN,, Dr. Terence Sim School of Computing National University of Singapore Outline 1 2 3 4 5 Outline 1 2 3 4 5 The Bayes Classifier is theoretically optimum. That is, prob. of error

More information

Distinguish between different types of scenes. Matching human perception Understanding the environment

Distinguish between different types of scenes. Matching human perception Understanding the environment Scene Recognition Adriana Kovashka UTCS, PhD student Problem Statement Distinguish between different types of scenes Applications Matching human perception Understanding the environment Indexing of images

More information

Elementary realization of BRST symmetry and gauge fixing

Elementary realization of BRST symmetry and gauge fixing Elementary realization of BRST symmetry and gauge fixing Martin Rocek, notes by Marcelo Disconzi Abstract This are notes from a talk given at Stony Brook University by Professor PhD Martin Rocek. I tried

More information

Vectors. January 13, 2013

Vectors. January 13, 2013 Vectors January 13, 2013 The simplest tensors are scalars, which are the measurable quantities of a theory, left invariant by symmetry transformations. By far the most common non-scalars are the vectors,

More information

Rotational Invariants for Wide-baseline Stereo

Rotational Invariants for Wide-baseline Stereo Rotational Invariants for Wide-baseline Stereo Jiří Matas, Petr Bílek, Ondřej Chum Centre for Machine Perception Czech Technical University, Department of Cybernetics Karlovo namesti 13, Prague, Czech

More information

Rigid Geometric Transformations

Rigid Geometric Transformations Rigid Geometric Transformations Carlo Tomasi This note is a quick refresher of the geometry of rigid transformations in three-dimensional space, expressed in Cartesian coordinates. 1 Cartesian Coordinates

More information

Gaussian discriminant analysis Naive Bayes

Gaussian discriminant analysis Naive Bayes DM825 Introduction to Machine Learning Lecture 7 Gaussian discriminant analysis Marco Chiarandini Department of Mathematics & Computer Science University of Southern Denmark Outline 1. is 2. Multi-variate

More information

Invariant Pattern Recognition using Dual-tree Complex Wavelets and Fourier Features

Invariant Pattern Recognition using Dual-tree Complex Wavelets and Fourier Features Invariant Pattern Recognition using Dual-tree Complex Wavelets and Fourier Features G. Y. Chen and B. Kégl Department of Computer Science and Operations Research, University of Montreal, CP 6128 succ.

More information

14 Singular Value Decomposition

14 Singular Value Decomposition 14 Singular Value Decomposition For any high-dimensional data analysis, one s first thought should often be: can I use an SVD? The singular value decomposition is an invaluable analysis tool for dealing

More information

Sparse representation classification and positive L1 minimization

Sparse representation classification and positive L1 minimization Sparse representation classification and positive L1 minimization Cencheng Shen Joint Work with Li Chen, Carey E. Priebe Applied Mathematics and Statistics Johns Hopkins University, August 5, 2014 Cencheng

More information

Gaussian and Linear Discriminant Analysis; Multiclass Classification

Gaussian and Linear Discriminant Analysis; Multiclass Classification Gaussian and Linear Discriminant Analysis; Multiclass Classification Professor Ameet Talwalkar Slide Credit: Professor Fei Sha Professor Ameet Talwalkar CS260 Machine Learning Algorithms October 13, 2015

More information

Math 462: Homework 2 Solutions

Math 462: Homework 2 Solutions Math 46: Homework Solutions Paul Hacking /6/. Consider the matrix A = (a) Check that A is orthogonal. (b) Determine whether A is a rotation or a reflection / rotary reflection. [Hint: What is the determinant

More information

Analysis of Hu's Moment Invariants on Image Scaling and Rotation

Analysis of Hu's Moment Invariants on Image Scaling and Rotation Edith Cowan University Research Online ECU Publications Pre. 11 1 Analysis of Hu's Moent Invariants on Iage Scaling and Rotation Zhihu Huang Edith Cowan University Jinsong Leng Edith Cowan University 1.119/ICCET.1.548554

More information

Homework 7-8 Solutions. Problems

Homework 7-8 Solutions. Problems Homework 7-8 Solutions Problems 26 A rhombus is a parallelogram with opposite sides of equal length Let us form a rhombus using vectors v 1 and v 2 as two adjacent sides, with v 1 = v 2 The diagonals of

More information

Lecture 05 Geometry of Least Squares

Lecture 05 Geometry of Least Squares Lecture 05 Geometry of Least Squares 16 September 2015 Taylor B. Arnold Yale Statistics STAT 312/612 Goals for today 1. Geometry of least squares 2. Projection matrix P and annihilator matrix M 3. Multivariate

More information

Solution to Problem Set No. 6: Time Independent Perturbation Theory

Solution to Problem Set No. 6: Time Independent Perturbation Theory Solution to Problem Set No. 6: Time Independent Perturbation Theory Simon Lin December, 17 1 The Anharmonic Oscillator 1.1 As a first step we invert the definitions of creation and annihilation operators

More information

Physics 342 Lecture 2. Linear Algebra I. Lecture 2. Physics 342 Quantum Mechanics I

Physics 342 Lecture 2. Linear Algebra I. Lecture 2. Physics 342 Quantum Mechanics I Physics 342 Lecture 2 Linear Algebra I Lecture 2 Physics 342 Quantum Mechanics I Wednesday, January 27th, 21 From separation of variables, we move to linear algebra Roughly speaking, this is the study

More information

19. TAYLOR SERIES AND TECHNIQUES

19. TAYLOR SERIES AND TECHNIQUES 19. TAYLOR SERIES AND TECHNIQUES Taylor polynomials can be generated for a given function through a certain linear combination of its derivatives. The idea is that we can approximate a function by a polynomial,

More information

Math 252 Fall 2002 Supplement on Euler s Method

Math 252 Fall 2002 Supplement on Euler s Method Math 5 Fall 00 Supplement on Euler s Method Introduction. The textbook seems overly enthusiastic about Euler s method. These notes aim to present a more realistic treatment of the value of the method and

More information

Generators for Continuous Coordinate Transformations

Generators for Continuous Coordinate Transformations Page 636 Lecture 37: Coordinate Transformations: Continuous Passive Coordinate Transformations Active Coordinate Transformations Date Revised: 2009/01/28 Date Given: 2009/01/26 Generators for Continuous

More information

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 18

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 18 EE/ACM 150 - Applications of Convex Optimization in Signal Processing and Communications Lecture 18 Andre Tkacenko Signal Processing Research Group Jet Propulsion Laboratory May 31, 2012 Andre Tkacenko

More information

G : Quantum Mechanics II

G : Quantum Mechanics II G5.666: Quantum Mechanics II Notes for Lecture 7 I. A SIMPLE EXAMPLE OF ANGULAR MOMENTUM ADDITION Given two spin-/ angular momenta, S and S, we define S S S The problem is to find the eigenstates of the

More information

EE100Su08 Lecture #11 (July 21 st 2008)

EE100Su08 Lecture #11 (July 21 st 2008) EE100Su08 Lecture #11 (July 21 st 2008) Bureaucratic Stuff Lecture videos should be up by tonight HW #2: Pick up from office hours today, will leave them in lab. REGRADE DEADLINE: Monday, July 28 th 2008,

More information

Math 320: Real Analysis MWF 1pm, Campion Hall 302 Homework 2 Solutions Please write neatly, and in complete sentences when possible.

Math 320: Real Analysis MWF 1pm, Campion Hall 302 Homework 2 Solutions Please write neatly, and in complete sentences when possible. Math 320: Real Analysis MWF 1pm, Campion Hall 302 Homework 2 Solutions Please write neatly, and in complete sentences when possible. Do the following problems from the book: 1.4.2, 1.4.4, 1.4.9, 1.4.11,

More information

Basic Concepts of. Feature Selection

Basic Concepts of. Feature Selection Basic Concepts of Pattern Recognition and Feature Selection Xiaojun Qi -- REU Site Program in CVMA (2011 Summer) 1 Outline Pattern Recognition Pattern vs. Features Pattern Classes Classification Feature

More information

ENGI 4430 PDEs - d Alembert Solutions Page 11.01

ENGI 4430 PDEs - d Alembert Solutions Page 11.01 ENGI 4430 PDEs - d Alembert Solutions Page 11.01 11. Partial Differential Equations Partial differential equations (PDEs) are equations involving functions of more than one variable and their partial derivatives

More information

General Relativity I

General Relativity I General Relativity I presented by John T. Whelan The University of Texas at Brownsville whelan@phys.utb.edu LIGO Livingston SURF Lecture 2002 July 5 General Relativity Lectures I. Today (JTW): Special

More information

26 Group Theory Basics

26 Group Theory Basics 26 Group Theory Basics 1. Reference: Group Theory and Quantum Mechanics by Michael Tinkham. 2. We said earlier that we will go looking for the set of operators that commute with the molecular Hamiltonian.

More information

Independent Component (IC) Models: New Extensions of the Multinormal Model

Independent Component (IC) Models: New Extensions of the Multinormal Model Independent Component (IC) Models: New Extensions of the Multinormal Model Davy Paindaveine (joint with Klaus Nordhausen, Hannu Oja, and Sara Taskinen) School of Public Health, ULB, April 2008 My research

More information

Introduction. Chapter Points, Vectors and Coordinate Systems

Introduction. Chapter Points, Vectors and Coordinate Systems Chapter 1 Introduction Computer aided geometric design (CAGD) concerns itself with the mathematical description of shape for use in computer graphics, manufacturing, or analysis. It draws upon the fields

More information

Representing regions in 2 ways:

Representing regions in 2 ways: Representing regions in 2 ways: Based on their external characteristics (its boundary): Shape characteristics Based on their internal characteristics (its region): Both Regional properties: color, texture,

More information

Discriminative Direction for Kernel Classifiers

Discriminative Direction for Kernel Classifiers Discriminative Direction for Kernel Classifiers Polina Golland Artificial Intelligence Lab Massachusetts Institute of Technology Cambridge, MA 02139 polina@ai.mit.edu Abstract In many scientific and engineering

More information

Math Homework 2

Math Homework 2 Math 73 Homework Due: September 8, 6 Suppose that f is holomorphic in a region Ω, ie an open connected set Prove that in any of the following cases (a) R(f) is constant; (b) I(f) is constant; (c) f is

More information

Image Normalization by Complex Moments

Image Normalization by Complex Moments 46 IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. PAMI-7, NO. 1, JANUARY 1985 Image Normalization by Complex Moments YASER S. ABU-MOSTAFA AND DEMETRI PSALTIS, MEMBER, IEEE Abstract-The

More information

1.1 A Scattering Experiment

1.1 A Scattering Experiment 1 Transfer Matrix In this chapter we introduce and discuss a mathematical method for the analysis of the wave propagation in one-dimensional systems. The method uses the transfer matrix and is commonly

More information

Low-Degree Polynomial Roots

Low-Degree Polynomial Roots Low-Degree Polynomial Roots David Eberly, Geometric Tools, Redmond WA 98052 https://www.geometrictools.com/ This work is licensed under the Creative Commons Attribution 4.0 International License. To view

More information

Lecture 8: Interest Point Detection. Saad J Bedros

Lecture 8: Interest Point Detection. Saad J Bedros #1 Lecture 8: Interest Point Detection Saad J Bedros sbedros@umn.edu Last Lecture : Edge Detection Preprocessing of image is desired to eliminate or at least minimize noise effects There is always tradeoff

More information

Equilibrium of a Rigid Body. Chapter 5

Equilibrium of a Rigid Body. Chapter 5 Equilibrium of a Rigid Body Chapter 5 Overview Rigid Body Equilibrium Free Body Diagrams Equations of Equilibrium 2 and 3-Force Members Statical Determinacy CONDITIONS FOR RIGID-BODY EQUILIBRIUM Recall

More information

Lecture 19 (Nov. 15, 2017)

Lecture 19 (Nov. 15, 2017) Lecture 19 8.31 Quantum Theory I, Fall 017 8 Lecture 19 Nov. 15, 017) 19.1 Rotations Recall that rotations are transformations of the form x i R ij x j using Einstein summation notation), where R is an

More information

Principal Component Analysis (PCA) CSC411/2515 Tutorial

Principal Component Analysis (PCA) CSC411/2515 Tutorial Principal Component Analysis (PCA) CSC411/2515 Tutorial Harris Chan Based on previous tutorial slides by Wenjie Luo, Ladislav Rampasek University of Toronto hchan@cs.toronto.edu October 19th, 2017 (UofT)

More information

9. Integral Ring Extensions

9. Integral Ring Extensions 80 Andreas Gathmann 9. Integral ing Extensions In this chapter we want to discuss a concept in commutative algebra that has its original motivation in algebra, but turns out to have surprisingly many applications

More information

MS 3011 Exercises. December 11, 2013

MS 3011 Exercises. December 11, 2013 MS 3011 Exercises December 11, 2013 The exercises are divided into (A) easy (B) medium and (C) hard. If you are particularly interested I also have some projects at the end which will deepen your understanding

More information

Learning Linear Detectors

Learning Linear Detectors Learning Linear Detectors Instructor - Simon Lucey 16-423 - Designing Computer Vision Apps Today Detection versus Classification Bayes Classifiers Linear Classifiers Examples of Detection 3 Learning: Detection

More information

AM 205: lecture 8. Last time: Cholesky factorization, QR factorization Today: how to compute the QR factorization, the Singular Value Decomposition

AM 205: lecture 8. Last time: Cholesky factorization, QR factorization Today: how to compute the QR factorization, the Singular Value Decomposition AM 205: lecture 8 Last time: Cholesky factorization, QR factorization Today: how to compute the QR factorization, the Singular Value Decomposition QR Factorization A matrix A R m n, m n, can be factorized

More information

GALOIS GROUPS OF CUBICS AND QUARTICS (NOT IN CHARACTERISTIC 2)

GALOIS GROUPS OF CUBICS AND QUARTICS (NOT IN CHARACTERISTIC 2) GALOIS GROUPS OF CUBICS AND QUARTICS (NOT IN CHARACTERISTIC 2) KEITH CONRAD We will describe a procedure for figuring out the Galois groups of separable irreducible polynomials in degrees 3 and 4 over

More information

2017 LCHL Paper 1 Table of Contents

2017 LCHL Paper 1 Table of Contents 3 7 10 2 2017 PAPER 1 INSTRUCTIONS There are two sections in this examination paper. Section A Concepts and Skills 150 marks 6 questions Section B Contexts and Applications 150 marks 3 questions Answer

More information

SWINGING UP A PENDULUM BY ENERGY CONTROL

SWINGING UP A PENDULUM BY ENERGY CONTROL Paper presented at IFAC 13th World Congress, San Francisco, California, 1996 SWINGING UP A PENDULUM BY ENERGY CONTROL K. J. Åström and K. Furuta Department of Automatic Control Lund Institute of Technology,

More information

Object Recognition Using Local Characterisation and Zernike Moments

Object Recognition Using Local Characterisation and Zernike Moments Object Recognition Using Local Characterisation and Zernike Moments A. Choksuriwong, H. Laurent, C. Rosenberger, and C. Maaoui Laboratoire Vision et Robotique - UPRES EA 2078, ENSI de Bourges - Université

More information

17.2 Nonhomogeneous Linear Equations. 27 September 2007

17.2 Nonhomogeneous Linear Equations. 27 September 2007 17.2 Nonhomogeneous Linear Equations 27 September 2007 Nonhomogeneous Linear Equations The differential equation to be studied is of the form ay (x) + by (x) + cy(x) = G(x) (1) where a 0, b, c are given

More information

Bonus Section II: Solving Trigonometric Equations

Bonus Section II: Solving Trigonometric Equations Fry Texas A&M University Math 150 Spring 2017 Bonus Section II 260 Bonus Section II: Solving Trigonometric Equations (In your text this section is found hiding at the end of 9.6) For what values of x does

More information

Sharp bounds on the VaR for sums of dependent risks

Sharp bounds on the VaR for sums of dependent risks Paul Embrechts Sharp bounds on the VaR for sums of dependent risks joint work with Giovanni Puccetti (university of Firenze, Italy) and Ludger Rüschendorf (university of Freiburg, Germany) Mathematical

More information