A Source Cell-phone Identification Scheme Based on Canonical Correlation Analysis of Photo Response Non-uniformity

Size: px
Start display at page:

Download "A Source Cell-phone Identification Scheme Based on Canonical Correlation Analysis of Photo Response Non-uniformity"

Transcription

1 Journal of Computational Information Systems 8: 4 (212) Available at A Source Cell-phone Identification Scheme Based on Canonical Analysis of Photo Response Non-uniformity Min LONG 1,, Ting PENG 1, Fei PENG 2 1 College of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha 41114, China 2 College of Information Science and Engineering, Hunan University, Changsha 4182, China Abstract According to characteristic of the photo response non-uniformity (PRNU) for the camera, a new method using canonical correlation analysis method to identify the source cell-phone of a photo is proposed. Compared with the typical method detecting a photo using a monochromatic channel, this method detecting a photo using three color channels can reflect the noise characteristics of imaging sensor more comprehensively. The experimental results demonstrate that the proposed method achieves improved performance. Keywords: Source Cell-phone Identification Forensic; Canonical Analysis; Imaging Sensor; PRNU Noise 1 Introduction With the rapid development of information technology, digital photos presented as evidence in a court is becoming possible, however, the authentication of them is still need to be resolved. Digital image forensics techniques are proposed to identify the source of images, distinguish synthetic images and real images, and determine whether the image has been tampered or not [1]. At present, source cell-phone identification are focused on researches by using pattern classifier to identify the source of photos with features extracted from the color, quality, wavelet coefficient statistics and binary similarity statistical characteristic of photos. In Reference [2], binary similarity features and image quality features are firstly extracted by using Sequential Forward Feature Selection (SFFS) algorithm, and then a Support Vector Machine (SVM) is trained to classify a group of 9 cell-phones. The identification accuracy is 83%, however, the existing methods can only identify the model of cell-phones, but can not identify the individual of them. 2 cell-phones with the same mode are tested with the method [3] and the identification accuracy is only 5%. Therefore, it s necessary to find an effective method to identify an individual cell-phone. Corresponding author. address: caslongm@yahoo.com.cn (Min LONG) / Copyright 212 Binary Information Press February 212

2 1434 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) It is known that cell-phone and camera have similar noise characteristics in imaging processing. Similarly, the individual cell-phone can be identified by its inherent fingerprint, namely PRNU noise produced by imaging sensor. Lukas firstly applied it for source camera identification [4], and it achieved a higher accuracy. Hu proposed a method for source camera identification based on large component of sensor pattern noise [5], however, only one color channel is employed. Since color images have three color channels, if they are analyzed simultaneously, it can reflect the noise characteristics of imaging sensor comprehensively. Based on the conception in Reference [5], a method is proposed in this paper to indentify the individual cell-phones by using canonical correlation analysis. First of all, three rough pattern noise are acquired by filter de-noising and de-cfa interpolation in three color channels of all photos, respectively, then the three reference pattern noises are obtained, finally the source cellphone of the photos are identified by using canonical correlation analysis method. 2 Noise in Cell-phone At present, CMOS(Complementary Metal Oxide Semiconductor) sensors are widely used in cellphones, which generate PRNU noise. [6]. PRNU noise is a multiplicative noise which changes along with the signal strength and it is caused by the inhomogenity of silicon wafers and imperfections during the sensor manufacturing process. i.e., different cell-phones of the same mode will have different PRNU noise. Therefore, the sources of photos captured from different cell-phones can be identified by analyzing their PRNU noise. Photo I stemmed from a cell-phone imaging pipeline can be defined as follows: I = F O + N (1) Where O denotes the ideal noise-free image, F and N denote the PRNU and the temporal noise, respectively. In general, P in Equation (2) is considered as the estimate of PRNU noise[7]. P = I f(i) (2) where f is a low-pass filter. The purpose of Equation (2) is to separate PRNU from image. Wiener filter [8] is a kind of linear filter based on the Minimum Mean Square Error (MMSE), which is widely used in image de-noising. Here, wiener filter is applied to extract PRNU. For multiple photos I i (i = 1, 2,, n) taken by a same cell-phone, an approximation can be obtained to form cell-phone pattern noise P cell by averaging their PRNU, which is formulated in Equation (3). P cell = 1 n n P i (3) i=1

3 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) Discription of the Identification Scheme 3.1 Canonical correlation analysis Canonical Analysis (CCA)was first proposed by Hotelling in [9]. The basic principle is described as follows: two representative variables are extracted by linear combination variables in two groups, respectively, and these two representative variables are used to reflect the correlation of two groups of variables. Given two zero mean random variables x R p and y R q, the target of CCA is to find a pair of direction a 1 and b 1, which obtain the maximum correlation between projection u 1 = a T 1 x and v 1 = b T 1 y, where u 1 and v 1 are named as the first pair canonical variables. The second pair u 2 and v 2 are formed which have the maximum correlation but are uncorrelated with u 1 and v 1, respectively. In a same way, L = min(p, q) pairs of canonical variables can be acquired. Assume X = [x 1, x 2...x N ] R p N and Y = [y 1, y 2...y N ] R q N are N samples of random variables x and y. The correlation coefficient ρ u,v is ρ u,v = E(uv) Eu 2 Ev 2 = E(a T xy T b) E(aT xx T a) E(b T yy T b) = a T C xy b at C xx ab T C yy b (4) where C xx R p p, C yy R q q, and they are the cross-covariance matrices of x and y, respectively. C xy R p q denotes their covariance matrix. Due to the extremum of ρ u,v has no relation with the magnitude of a and b, except the direction of them, a constraint condition is defined as shown in Equation (5). a T C xx a = b T C yy b = 1 (5) Therefore, the objective of CAA is to compute the maximum of ρ u,v with the constrain of Equation (5), i.e., ρ max = a T C xy b (6) Applying Lagrange multiplier technique to Equation (6), then get L(a, b) = a T C xy b λ 1 2 (at C xx a 1) λ 2 2 (bt C yy b 1) (7) where λ 1 and λ 2 are Lagrange multipliers. Partial derivatives are done to a and b, respectively, L a = C xyb λ 1 C xx a = (8) L b = C yxa λ 2 C yy b = (9) a T and b T are multiplied to Equation (8) and Equation (9), respectively, a T C xy b = a T λ 1 C xx a (1) b T C yx a = b T λ 2 C yy b (11)

4 1436 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) Obviously, λ 1 = λ T 2 = λ 2. Given λ = λ 1 = λ 2, according to Equation (6), λ = ρ max is obtained. Inserting λ into Equation (8) and Equation (9), ( Cxy C yx ) ( a ) b = λ ( Cxx C xy ) ( a ) b (12) Eigenvectors a, b and eigenvalue λ can be computed according to Equation (12). 3.2 Identification scheme of source cell-phone The proposed scheme is described as follows: Step1 Extraction of cell-phone pattern noise a) As for all photos I i (i = 1, 2...n), noise extraction and de-cfa interpolation operation are done to in each color channel (including R, G, B), then these noises are averaged as three rough reference pattern noises W r, W g,and W b, respectively. b) W r, W g, and W b are sorted in descending order, and the first n elements are converted to three row vectors P r, P g, and P b, respectively. Here n is 5% of the number of pixels. Step2 Identification process To identify whether a special photo J was taken by cell-phone M, three noises N r, N g and N b from R, G, B channels of the photo are extracted through the processing similar to step1, and the corresponding row vector n r, n g, n b are acquired according to step1-b), then get x = (p r, p g, p b ) T and y = (n r, n g, n b ) T. Since noise usually follows the hypothesis of zero mean Gaussian distribution and the correlation ρ between X = [x 1, x 2...x N ] R p N and Y = [y 1, y 2...y N ] R q N can be calculated according to Equation (4). Generally, the correlations between noises of photos taken by a same cell-phone should be strong, otherwise the correlations should be weak. Therefore, the source cell-phone of a photo can be identified by comparing ρ by setting a proper threshold T as shown in Equation (13). ρ < T, { ρ > T, photo J was taken by cell - phone M photo J wasn t taken by cell - phone M (13) Here, the threshold T is determined in experiments. 4 Experiment Four cell-phones are selected for experiments, and 3 photos are taken from each cell-phone, where 15 photos are used for extracting cell-phone pattern noise and the left 15 photos are used for identification tests. The model of the cell-phones are shown in Table 1. The resolutions of all photos are adjusted to According to Reference [5], the smaller n is, the stronger the correlations are, but the standard deviation will increase when the correlation mean increase, which will result in the increases of

5 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) Table 1: The cell-phones used in the experiments Cell-phone model Sensor type The largest resolution CMOS CMOS CMOS CMOS detection error. In order to compromise the influence of mean and standard deviation, the value of n is selected as 5% of the whole number of pixels in experiments. The comparison results of correlation mean and standard deviation of two methods are shown in Table 2. Table 2: mean and standard deviation of two methods Cell-phone model Method in [5] Our method mean standard deviation mean standard deviation As seen in Table 2, the proposed method has little improvement to the correlation standard deviation, but has significant impact on the correlation mean. That is, the distance between photos with matching cell-phone and mismatching cell phones is enlarged. Here, canonical variables are set as p = q = 3, L = 3. According to the principle of CCA, ρ have three values: ρ 1, ρ 2 and ρ 3, where ρ 1 > ρ 2 > ρ 3. when calculating canonical correlation, it is processed as follows: ρ = ρ 3, { ρ1, when a photo matches the cell-phone when a photo dismatches the cell-phone (14) Experiments are done to the proposed method and the method in Reference [5] to analysis their performance, and the results are shown in Fig. 1 Fig. 4. It can be seen that the dispersion degree of the proposed method is smaller than that of the method in Reference [5], which will improve the accuracy of identification. According to Neyman-Pearson criterion, a threshold is calculated to minimize the false rejection rate (FRR) with a given false acceptance rate (FAR). The threshold T and FRR of all 4 cell-phones for the method in Reference [5] and the proposed method is shown in Table 3, where FAR=.1. It is can be seen that the proposed method can achieve a higher accuracy of identification than that of the method in Reference [5].

6 1438 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) Table 3: mean and standard deviation of two methods Cell-phone model Method in [5] Our method T FRR T FRR (a) Experiment results of the proposed method (b) Experiment results of the method in Reference [5] Fig. 1: Distribution of correlations between photos taken by 4 cell-phones and pattern noise from (Each type has 15 photos) (a) Experiment results of the proposed method (b) Experiment results of the method in Reference [5] Fig. 2: Distribution of correlations between photos taken by 4 cell-phones and pattern noise from (Each type has 15 photos).

7 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) (a) Experiment results of the proposed method (b) Experiment results of the method in Reference [5] Fig. 3: Distribution of correlations between photos taken by 4 cell-phones and pattern noise from (Each type has 15 photos) (a) Experiment results of the proposed method (b) Experiment results of the method in Reference [5] Fig. 4: Distribution of correlations between photos taken by 4 cell-phones and pattern noise from (Each type has 15 photos). 5 Conclusion In this paper, a method using CCA to detect the source cell-phones of photos is proposed. Compared with the method in Reference [5], its advantage lies in breaking the limitations of detecting in one color channel, and three color channels are all used in the proposed method simultaneously, which makes the result more reliable. Moreover, correlation maximization of CCA can distinguish photos from different cell phones more effectively. The experimental results also illustrate its feasibility. It has great potential in the application of distinguish the source of photos from different cell-phones.

8 144 M. Long et al. /Journal of Computational Information Systems 8: 4 (212) Acknowledgement This work is supported by the Network and Information Security Key Laboratory Foundation of Hunan Province, China (Grant No. NISL212), and the National Natural Science Foundation of China (Grant No. 6114, No ) and the Education Department Foundation of Hunan Province (Grant No. 11B2). References [1] H. T. Sencar and N. Memon.Overview of State-of-the-Art in digital image forensics. In Indian Statistical Institute Platinum Jubilee Monograph series titled Statistical Science and Interdisciplinary Research. World Scientific, 28. [2] O. Celiktutan, B. Sankur, I. Avcıbas, and N. Memon. Source cell-phone identification. Proc. AD- COM, pages 1 3, 26. [3] V. T. Lanh, S. Emmanuel, and M. S. Kankanhalli. Identifying source cell phone using chromatic aberration. In Proc. IEEE Conference on Multimedia and Expo, pages 2 5, 27. [4] J. Lukas, J. Fridrich, and M. Goljan. Digital camera identification from sensor pattern noise. IEEE Trans.Inf. Forensics Security, pages , 26. [5] Yongjian Hu, Binghua Yu, and Chao Jian. Source camera identification using large components of sensor pattern noise. computer applications, pages 31 35, 21. [6] A. El Gamal and H. Eltoukhy. CMOS image sensors. IEEE Circuits and Devices Magazine, pages 6 2, 25. [7] Erwin J. Alles, Zeno J. M. H. Geradts, and Cor J. Veenman. Source camera identification for low resolution heavily compressed images. In Proc. of IEEE International Conference on Computational Sciences and its Applications, pages , 28. [8] A. Castiglione, G. Cattaneo, M. Cembalo, and U. F. Petrillo. Source camera identification in real practice: a preliminary experimentation. In Proc.of IEEE Conference on BWCCA, pages , 21. [9] H. hotelling. Relations between two sets of variates. Biometrika, pages , 1936.

Putting the PRNU Model in Reverse Gear: Findings with Synthetic Signals

Putting the PRNU Model in Reverse Gear: Findings with Synthetic Signals Putting the PRNU Model in Reverse Gear: Findings with Synthetic Signals Miguel Masciopinto, Fernando Pérez-González Signal Theory and Communications Department, University of Vigo E. E. Telecomunicación,

More information

Design of Projection Matrices for PRNU Compression

Design of Projection Matrices for PRNU Compression Design of Projection Matrices for PRNU Compression Luca Bondi 1, Fernando Pérez-González 2, Paolo Bestagini 1 and Stefano Tubaro 1 1 Dipartimento di Informazione, Elettronica e Bioingegneria, Politecnico

More information

Canonical Correlation Analysis of Longitudinal Data

Canonical Correlation Analysis of Longitudinal Data Biometrics Section JSM 2008 Canonical Correlation Analysis of Longitudinal Data Jayesh Srivastava Dayanand N Naik Abstract Studying the relationship between two sets of variables is an important multivariate

More information

Boosting: Algorithms and Applications

Boosting: Algorithms and Applications Boosting: Algorithms and Applications Lecture 11, ENGN 4522/6520, Statistical Pattern Recognition and Its Applications in Computer Vision ANU 2 nd Semester, 2008 Chunhua Shen, NICTA/RSISE Boosting Definition

More information

Individual Camera Device Identification from JPEG Images

Individual Camera Device Identification from JPEG Images Signal Processing: Image Communication (6) 5 Signal Processing: Image Communication Individual Camera Device Identification from JPEG Images Tong Qiao a,, Florent Retraint, Rémi Cogranne a, Thanh Hai Thai

More information

Compressive Sensing Forensics Xiaoyu Chu, Student Member, IEEE, Matthew Christopher Stamm, Member, IEEE, and K. J. Ray Liu, Fellow, IEEE

Compressive Sensing Forensics Xiaoyu Chu, Student Member, IEEE, Matthew Christopher Stamm, Member, IEEE, and K. J. Ray Liu, Fellow, IEEE 1416 IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, VOL. 10, NO. 7, JULY 2015 Compressive Sensing Forensics Xiaoyu Chu, Student Member, IEEE, Matthew Christopher Stamm, Member, IEEE, and K. J.

More information

Multimedia communications

Multimedia communications Multimedia communications Comunicazione multimediale G. Menegaz gloria.menegaz@univr.it Prologue Context Context Scale Scale Scale Course overview Goal The course is about wavelets and multiresolution

More information

Maximum variance formulation

Maximum variance formulation 12.1. Principal Component Analysis 561 Figure 12.2 Principal component analysis seeks a space of lower dimensionality, known as the principal subspace and denoted by the magenta line, such that the orthogonal

More information

Edges and Scale. Image Features. Detecting edges. Origin of Edges. Solution: smooth first. Effects of noise

Edges and Scale. Image Features. Detecting edges. Origin of Edges. Solution: smooth first. Effects of noise Edges and Scale Image Features From Sandlot Science Slides revised from S. Seitz, R. Szeliski, S. Lazebnik, etc. Origin of Edges surface normal discontinuity depth discontinuity surface color discontinuity

More information

Managing a Large Database of Camera Fingerprints

Managing a Large Database of Camera Fingerprints Managing a Large Database of Camera Fingerprints Miroslav Goljan, Jessica Fridrich, Tomáš Filler Department of Electrical and Computer Engineering State University of New Yor, Binghamton, New Yor, 390-6000

More information

The Method of Obtaining Best Unary Polynomial for the Chaotic Sequence of Image Encryption

The Method of Obtaining Best Unary Polynomial for the Chaotic Sequence of Image Encryption Journal of Information Hiding and Multimedia Signal Processing c 2017 ISSN 2073-4212 Ubiquitous International Volume 8, Number 5, September 2017 The Method of Obtaining Best Unary Polynomial for the Chaotic

More information

Canonical Correlations

Canonical Correlations Canonical Correlations Like Principal Components Analysis, Canonical Correlation Analysis looks for interesting linear combinations of multivariate observations. In Canonical Correlation Analysis, a multivariate

More information

Reliability of Seismic Data for Hydrocarbon Reservoir Characterization

Reliability of Seismic Data for Hydrocarbon Reservoir Characterization Reliability of Seismic Data for Hydrocarbon Reservoir Characterization Geetartha Dutta (gdutta@stanford.edu) December 10, 2015 Abstract Seismic data helps in better characterization of hydrocarbon reservoirs.

More information

ELEG 5633 Detection and Estimation Signal Detection: Deterministic Signals

ELEG 5633 Detection and Estimation Signal Detection: Deterministic Signals ELEG 5633 Detection and Estimation Signal Detection: Deterministic Signals Jingxian Wu Department of Electrical Engineering University of Arkansas Outline Matched Filter Generalized Matched Filter Signal

More information

1 Principal Components Analysis

1 Principal Components Analysis Lecture 3 and 4 Sept. 18 and Sept.20-2006 Data Visualization STAT 442 / 890, CM 462 Lecture: Ali Ghodsi 1 Principal Components Analysis Principal components analysis (PCA) is a very popular technique for

More information

Ch.3 Canonical correlation analysis (CCA) [Book, Sect. 2.4]

Ch.3 Canonical correlation analysis (CCA) [Book, Sect. 2.4] Ch.3 Canonical correlation analysis (CCA) [Book, Sect. 2.4] With 2 sets of variables {x i } and {y j }, canonical correlation analysis (CCA), first introduced by Hotelling (1936), finds the linear modes

More information

I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Canonical Edps/Soc 584 and Psych 594 Applied Multivariate Statistics Carolyn J. Anderson Department of Educational Psychology I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Canonical Slide

More information

Wavelet Packet Based Digital Image Watermarking

Wavelet Packet Based Digital Image Watermarking Wavelet Packet Based Digital Image ing A.Adhipathi Reddy, B.N.Chatterji Department of Electronics and Electrical Communication Engg. Indian Institute of Technology, Kharagpur 72 32 {aar, bnc}@ece.iitkgp.ernet.in

More information

Image Recognition Using Modified Zernike Moments

Image Recognition Using Modified Zernike Moments Sensors & Transducers 204 by IFSA Publishing S. L. http://www.sensorsportal.com Image Recognition Using Modified ernike Moments Min HUANG Yaqiong MA and Qiuping GONG School of Computer and Communication

More information

Auxiliaire d enseignement Nicolas Ayotte

Auxiliaire d enseignement Nicolas Ayotte 2012-02-15 GEL 4203 / GEL 7041 OPTOÉLECTRONIQUE Auxiliaire d enseignement Nicolas Ayotte GEL 4203 / GEL 7041 Optoélectronique VI PN JUNCTION The density of charge sign Fixed charge density remaining 2

More information

CITS 4402 Computer Vision

CITS 4402 Computer Vision CITS 4402 Computer Vision A/Prof Ajmal Mian Adj/A/Prof Mehdi Ravanbakhsh Lecture 06 Object Recognition Objectives To understand the concept of image based object recognition To learn how to match images

More information

Chapter 2 Canonical Correlation Analysis

Chapter 2 Canonical Correlation Analysis Chapter 2 Canonical Correlation Analysis Canonical correlation analysis CCA, which is a multivariate analysis method, tries to quantify the amount of linear relationships etween two sets of random variales,

More information

Unsupervised Machine Learning and Data Mining. DS 5230 / DS Fall Lecture 7. Jan-Willem van de Meent

Unsupervised Machine Learning and Data Mining. DS 5230 / DS Fall Lecture 7. Jan-Willem van de Meent Unsupervised Machine Learning and Data Mining DS 5230 / DS 4420 - Fall 2018 Lecture 7 Jan-Willem van de Meent DIMENSIONALITY REDUCTION Borrowing from: Percy Liang (Stanford) Dimensionality Reduction Goal:

More information

Introduction. x 1 x 2. x n. y 1

Introduction. x 1 x 2. x n. y 1 This article, an update to an original article by R. L. Malacarne, performs a canonical correlation analysis on financial data of country - specific Exchange Traded Funds (ETFs) to analyze the relationship

More information

7. Variable extraction and dimensionality reduction

7. Variable extraction and dimensionality reduction 7. Variable extraction and dimensionality reduction The goal of the variable selection in the preceding chapter was to find least useful variables so that it would be possible to reduce the dimensionality

More information

Machine Learning 2nd Edition

Machine Learning 2nd Edition INTRODUCTION TO Lecture Slides for Machine Learning 2nd Edition ETHEM ALPAYDIN, modified by Leonardo Bobadilla and some parts from http://www.cs.tau.ac.il/~apartzin/machinelearning/ The MIT Press, 2010

More information

Drift Reduction For Metal-Oxide Sensor Arrays Using Canonical Correlation Regression And Partial Least Squares

Drift Reduction For Metal-Oxide Sensor Arrays Using Canonical Correlation Regression And Partial Least Squares Drift Reduction For Metal-Oxide Sensor Arrays Using Canonical Correlation Regression And Partial Least Squares R Gutierrez-Osuna Computer Science Department, Wright State University, Dayton, OH 45435,

More information

What is Image Deblurring?

What is Image Deblurring? What is Image Deblurring? When we use a camera, we want the recorded image to be a faithful representation of the scene that we see but every image is more or less blurry, depending on the circumstances.

More information

A Laplacian of Gaussian-based Approach for Spot Detection in Two-Dimensional Gel Electrophoresis Images

A Laplacian of Gaussian-based Approach for Spot Detection in Two-Dimensional Gel Electrophoresis Images A Laplacian of Gaussian-based Approach for Spot Detection in Two-Dimensional Gel Electrophoresis Images Feng He 1, Bangshu Xiong 1, Chengli Sun 1, Xiaobin Xia 1 1 Key Laboratory of Nondestructive Test

More information

2. What are the tradeoffs among different measures of error (e.g. probability of false alarm, probability of miss, etc.)?

2. What are the tradeoffs among different measures of error (e.g. probability of false alarm, probability of miss, etc.)? ECE 830 / CS 76 Spring 06 Instructors: R. Willett & R. Nowak Lecture 3: Likelihood ratio tests, Neyman-Pearson detectors, ROC curves, and sufficient statistics Executive summary In the last lecture we

More information

A Least Squares Formulation for Canonical Correlation Analysis

A Least Squares Formulation for Canonical Correlation Analysis A Least Squares Formulation for Canonical Correlation Analysis Liang Sun, Shuiwang Ji, and Jieping Ye Department of Computer Science and Engineering Arizona State University Motivation Canonical Correlation

More information

Biometrics: Introduction and Examples. Raymond Veldhuis

Biometrics: Introduction and Examples. Raymond Veldhuis Biometrics: Introduction and Examples Raymond Veldhuis 1 Overview Biometric recognition Face recognition Challenges Transparent face recognition Large-scale identification Watch list Anonymous biometrics

More information

COMP 551 Applied Machine Learning Lecture 13: Dimension reduction and feature selection

COMP 551 Applied Machine Learning Lecture 13: Dimension reduction and feature selection COMP 551 Applied Machine Learning Lecture 13: Dimension reduction and feature selection Instructor: Herke van Hoof (herke.vanhoof@cs.mcgill.ca) Based on slides by:, Jackie Chi Kit Cheung Class web page:

More information

Lecture 5: Likelihood ratio tests, Neyman-Pearson detectors, ROC curves, and sufficient statistics. 1 Executive summary

Lecture 5: Likelihood ratio tests, Neyman-Pearson detectors, ROC curves, and sufficient statistics. 1 Executive summary ECE 830 Spring 207 Instructor: R. Willett Lecture 5: Likelihood ratio tests, Neyman-Pearson detectors, ROC curves, and sufficient statistics Executive summary In the last lecture we saw that the likelihood

More information

Image Acquisition and Sampling Theory

Image Acquisition and Sampling Theory Image Acquisition and Sampling Theory Electromagnetic Spectrum The wavelength required to see an object must be the same size of smaller than the object 2 Image Sensors 3 Sensor Strips 4 Digital Image

More information

Role of Assembling Invariant Moments and SVM in Fingerprint Recognition

Role of Assembling Invariant Moments and SVM in Fingerprint Recognition 56 Role of Assembling Invariant Moments SVM in Fingerprint Recognition 1 Supriya Wable, 2 Chaitali Laulkar 1, 2 Department of Computer Engineering, University of Pune Sinhgad College of Engineering, Pune-411

More information

Introduction p. 1 Fundamental Problems p. 2 Core of Fundamental Theory and General Mathematical Ideas p. 3 Classical Statistical Decision p.

Introduction p. 1 Fundamental Problems p. 2 Core of Fundamental Theory and General Mathematical Ideas p. 3 Classical Statistical Decision p. Preface p. xiii Acknowledgment p. xix Introduction p. 1 Fundamental Problems p. 2 Core of Fundamental Theory and General Mathematical Ideas p. 3 Classical Statistical Decision p. 4 Bayes Decision p. 5

More information

Robust extraction of specific signals with temporal structure

Robust extraction of specific signals with temporal structure Robust extraction of specific signals with temporal structure Zhi-Lin Zhang, Zhang Yi Computational Intelligence Laboratory, School of Computer Science and Engineering, University of Electronic Science

More information

2D Image Processing Face Detection and Recognition

2D Image Processing Face Detection and Recognition 2D Image Processing Face Detection and Recognition Prof. Didier Stricker Kaiserlautern University http://ags.cs.uni-kl.de/ DFKI Deutsches Forschungszentrum für Künstliche Intelligenz http://av.dfki.de

More information

Orientation Map Based Palmprint Recognition

Orientation Map Based Palmprint Recognition Orientation Map Based Palmprint Recognition (BM) 45 Orientation Map Based Palmprint Recognition B. H. Shekar, N. Harivinod bhshekar@gmail.com, harivinodn@gmail.com India, Mangalore University, Department

More information

Data Mining Techniques

Data Mining Techniques Data Mining Techniques CS 6220 - Section 3 - Fall 2016 Lecture 12 Jan-Willem van de Meent (credit: Yijun Zhao, Percy Liang) DIMENSIONALITY REDUCTION Borrowing from: Percy Liang (Stanford) Linear Dimensionality

More information

Pattern Recognition and Machine Learning

Pattern Recognition and Machine Learning Christopher M. Bishop Pattern Recognition and Machine Learning ÖSpri inger Contents Preface Mathematical notation Contents vii xi xiii 1 Introduction 1 1.1 Example: Polynomial Curve Fitting 4 1.2 Probability

More information

ESANN'2001 proceedings - European Symposium on Artificial Neural Networks Bruges (Belgium), April 2001, D-Facto public., ISBN ,

ESANN'2001 proceedings - European Symposium on Artificial Neural Networks Bruges (Belgium), April 2001, D-Facto public., ISBN , Sparse Kernel Canonical Correlation Analysis Lili Tan and Colin Fyfe 2, Λ. Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong. 2. School of Information and Communication

More information

IN Pratical guidelines for classification Evaluation Feature selection Principal component transform Anne Solberg

IN Pratical guidelines for classification Evaluation Feature selection Principal component transform Anne Solberg IN 5520 30.10.18 Pratical guidelines for classification Evaluation Feature selection Principal component transform Anne Solberg (anne@ifi.uio.no) 30.10.18 IN 5520 1 Literature Practical guidelines of classification

More information

Finding the best mismatched detector for channel coding and hypothesis testing

Finding the best mismatched detector for channel coding and hypothesis testing Finding the best mismatched detector for channel coding and hypothesis testing Sean Meyn Department of Electrical and Computer Engineering University of Illinois and the Coordinated Science Laboratory

More information

Feature extraction: Corners and blobs

Feature extraction: Corners and blobs Feature extraction: Corners and blobs Review: Linear filtering and edge detection Name two different kinds of image noise Name a non-linear smoothing filter What advantages does median filtering have over

More information

INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY

INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY [Gaurav, 2(1): Jan., 2013] ISSN: 2277-9655 IJESRT INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY Face Identification & Detection Using Eigenfaces Sachin.S.Gurav *1, K.R.Desai 2 *1

More information

Robotics 2 Target Tracking. Kai Arras, Cyrill Stachniss, Maren Bennewitz, Wolfram Burgard

Robotics 2 Target Tracking. Kai Arras, Cyrill Stachniss, Maren Bennewitz, Wolfram Burgard Robotics 2 Target Tracking Kai Arras, Cyrill Stachniss, Maren Bennewitz, Wolfram Burgard Slides by Kai Arras, Gian Diego Tipaldi, v.1.1, Jan 2012 Chapter Contents Target Tracking Overview Applications

More information

Microphone Identification using Higher-Order Statistics

Microphone Identification using Higher-Order Statistics Microphone Identification using Higher-Order Statistics Hafiz Malik 1, and John W. Miller 1 1 Information Systems, Security, and Forensics Lab, Department of Electrical & Computer Engineering, University

More information

Machine Learning for Software Engineering

Machine Learning for Software Engineering Machine Learning for Software Engineering Dimensionality Reduction Prof. Dr.-Ing. Norbert Siegmund Intelligent Software Systems 1 2 Exam Info Scheduled for Tuesday 25 th of July 11-13h (same time as the

More information

Novel spectrum sensing schemes for Cognitive Radio Networks

Novel spectrum sensing schemes for Cognitive Radio Networks Novel spectrum sensing schemes for Cognitive Radio Networks Cantabria University Santander, May, 2015 Supélec, SCEE Rennes, France 1 The Advanced Signal Processing Group http://gtas.unican.es The Advanced

More information

Biometric Security Based on ECG

Biometric Security Based on ECG Biometric Security Based on ECG Lingni Ma, J.A. de Groot and Jean-Paul Linnartz Eindhoven University of Technology Signal Processing Systems, Electrical Engineering l.ma.1@student.tue.nl j.a.d.groot@tue.nl

More information

Introduction to Machine Learning

Introduction to Machine Learning 10-701 Introduction to Machine Learning PCA Slides based on 18-661 Fall 2018 PCA Raw data can be Complex, High-dimensional To understand a phenomenon we measure various related quantities If we knew what

More information

GEOG 4110/5100 Advanced Remote Sensing Lecture 15

GEOG 4110/5100 Advanced Remote Sensing Lecture 15 GEOG 4110/5100 Advanced Remote Sensing Lecture 15 Principal Component Analysis Relevant reading: Richards. Chapters 6.3* http://www.ce.yildiz.edu.tr/personal/songul/file/1097/principal_components.pdf *For

More information

Kernel-Based Contrast Functions for Sufficient Dimension Reduction

Kernel-Based Contrast Functions for Sufficient Dimension Reduction Kernel-Based Contrast Functions for Sufficient Dimension Reduction Michael I. Jordan Departments of Statistics and EECS University of California, Berkeley Joint work with Kenji Fukumizu and Francis Bach

More information

MultiDimensional Signal Processing Master Degree in Ingegneria delle Telecomunicazioni A.A

MultiDimensional Signal Processing Master Degree in Ingegneria delle Telecomunicazioni A.A MultiDimensional Signal Processing Master Degree in Ingegneria delle Telecomunicazioni A.A. 2017-2018 Pietro Guccione, PhD DEI - DIPARTIMENTO DI INGEGNERIA ELETTRICA E DELL INFORMAZIONE POLITECNICO DI

More information

12.4 Known Channel (Water-Filling Solution)

12.4 Known Channel (Water-Filling Solution) ECEn 665: Antennas and Propagation for Wireless Communications 54 2.4 Known Channel (Water-Filling Solution) The channel scenarios we have looed at above represent special cases for which the capacity

More information

Watermark Extraction Optimization Using PSO Algorithm

Watermark Extraction Optimization Using PSO Algorithm Research Journal of Applied Sciences, Engineering and Technology 5(12): 3312-3319, 2013 ISS: 2040-7459; e-iss: 2040-7467 Maxwell Scientific Organization, 2013 Submitted: June 09, 2012 Accepted: July 18,

More information

Atomic Resolution Interfacial Structure of Lead-free Ferroelectric

Atomic Resolution Interfacial Structure of Lead-free Ferroelectric Atomic Resolution Interfacial Structure of Lead-free Ferroelectric K 0.5 Na 0.5 NbO 3 Thin films Deposited on SrTiO 3 Chao Li 1, Lingyan Wang 1*, Zhao Wang 2, Yaodong Yang 2, Wei Ren 1 and Guang Yang 1

More information

Visualization of polarization state and its application in Optics classroom teaching

Visualization of polarization state and its application in Optics classroom teaching Visualization of polarization state and its application in Optics classroom teaching Bing Lei 1, *, Wei Liu 1, Jianhua Shi 1, Wei Wang 1, Tianfu Yao 1 and Shugang Liu 2 1 College of Optoelectronic Science

More information

Face detection and recognition. Detection Recognition Sally

Face detection and recognition. Detection Recognition Sally Face detection and recognition Detection Recognition Sally Face detection & recognition Viola & Jones detector Available in open CV Face recognition Eigenfaces for face recognition Metric learning identification

More information

Over-enhancement Reduction in Local Histogram Equalization using its Degrees of Freedom. Alireza Avanaki

Over-enhancement Reduction in Local Histogram Equalization using its Degrees of Freedom. Alireza Avanaki Over-enhancement Reduction in Local Histogram Equalization using its Degrees of Freedom Alireza Avanaki ABSTRACT A well-known issue of local (adaptive) histogram equalization (LHE) is over-enhancement

More information

Automatic Identity Verification Using Face Images

Automatic Identity Verification Using Face Images Automatic Identity Verification Using Face Images Sabry F. Saraya and John F. W. Zaki Computer & Systems Eng. Dept. Faculty of Engineering Mansoura University. Abstract This paper presents two types of

More information

arxiv: v2 [cs.cr] 6 Aug 2017

arxiv: v2 [cs.cr] 6 Aug 2017 Cryptanalyzing an Image Scrambling Encryption Algorithm of Pixel Bits Chengqing Li a,, Dongdong Lin a, Jinhu Lü b a Hunan Province Cooperative Innovation Center for Wind Power Equipment and Energy Conversion,

More information

UCSD ECE153 Handout #30 Prof. Young-Han Kim Thursday, May 15, Homework Set #6 Due: Thursday, May 22, 2011

UCSD ECE153 Handout #30 Prof. Young-Han Kim Thursday, May 15, Homework Set #6 Due: Thursday, May 22, 2011 UCSD ECE153 Handout #30 Prof. Young-Han Kim Thursday, May 15, 2014 Homework Set #6 Due: Thursday, May 22, 2011 1. Linear estimator. Consider a channel with the observation Y = XZ, where the signal X and

More information

CHAPTER 4 PRINCIPAL COMPONENT ANALYSIS-BASED FUSION

CHAPTER 4 PRINCIPAL COMPONENT ANALYSIS-BASED FUSION 59 CHAPTER 4 PRINCIPAL COMPONENT ANALYSIS-BASED FUSION 4. INTRODUCTION Weighted average-based fusion algorithms are one of the widely used fusion methods for multi-sensor data integration. These methods

More information

Controlling the Period-Doubling Bifurcation of Logistic Model

Controlling the Period-Doubling Bifurcation of Logistic Model ISSN 1749-3889 (print), 1749-3897 (online) International Journal of Nonlinear Science Vol.20(2015) No.3,pp.174-178 Controlling the Period-Doubling Bifurcation of Logistic Model Zhiqian Wang 1, Jiashi Tang

More information

The Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment

The Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment he Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment William Glunt 1, homas L. Hayden 2 and Robert Reams 2 1 Department of Mathematics and Computer Science, Austin Peay State

More information

Myoelectrical signal classification based on S transform and two-directional 2DPCA

Myoelectrical signal classification based on S transform and two-directional 2DPCA Myoelectrical signal classification based on S transform and two-directional 2DPCA Hong-Bo Xie1 * and Hui Liu2 1 ARC Centre of Excellence for Mathematical and Statistical Frontiers Queensland University

More information

Laser on-line Thickness Measurement Technology Based on Judgment and Wavelet De-noising

Laser on-line Thickness Measurement Technology Based on Judgment and Wavelet De-noising Sensors & Transducers, Vol. 168, Issue 4, April 214, pp. 137-141 Sensors & Transducers 214 by IFSA Publishing, S. L. http://www.sensorsportal.com Laser on-line Thickness Measurement Technology Based on

More information

A Statistical Analysis of Fukunaga Koontz Transform

A Statistical Analysis of Fukunaga Koontz Transform 1 A Statistical Analysis of Fukunaga Koontz Transform Xiaoming Huo Dr. Xiaoming Huo is an assistant professor at the School of Industrial and System Engineering of the Georgia Institute of Technology,

More information

Underwater Target Detection from Multi-Platform Sonar Imagery Using Multi-Channel Coherence Analysis

Underwater Target Detection from Multi-Platform Sonar Imagery Using Multi-Channel Coherence Analysis Proceedings of the 29 IEEE International Conference on Systems, Man, and Cybernetics San Antonio, TX, USA - October 29 Underwater Target Detection from Multi-Platform Sonar Imagery Using Multi-Channel

More information

Shannon meets Wiener II: On MMSE estimation in successive decoding schemes

Shannon meets Wiener II: On MMSE estimation in successive decoding schemes Shannon meets Wiener II: On MMSE estimation in successive decoding schemes G. David Forney, Jr. MIT Cambridge, MA 0239 USA forneyd@comcast.net Abstract We continue to discuss why MMSE estimation arises

More information

Waveform-Based Coding: Outline

Waveform-Based Coding: Outline Waveform-Based Coding: Transform and Predictive Coding Yao Wang Polytechnic University, Brooklyn, NY11201 http://eeweb.poly.edu/~yao Based on: Y. Wang, J. Ostermann, and Y.-Q. Zhang, Video Processing and

More information

6.867 Machine Learning

6.867 Machine Learning 6.867 Machine Learning Problem Set 2 Due date: Wednesday October 6 Please address all questions and comments about this problem set to 6867-staff@csail.mit.edu. You will need to use MATLAB for some of

More information

Learning SVM Classifiers with Indefinite Kernels

Learning SVM Classifiers with Indefinite Kernels Learning SVM Classifiers with Indefinite Kernels Suicheng Gu and Yuhong Guo Dept. of Computer and Information Sciences Temple University Support Vector Machines (SVMs) (Kernel) SVMs are widely used in

More information

A Systematic Description of Source Significance Information

A Systematic Description of Source Significance Information A Systematic Description of Source Significance Information Norbert Goertz Institute for Digital Communications School of Engineering and Electronics The University of Edinburgh Mayfield Rd., Edinburgh

More information

Old painting digital color restoration

Old painting digital color restoration Old painting digital color restoration Michail Pappas Ioannis Pitas Dept. of Informatics, Aristotle University of Thessaloniki GR-54643 Thessaloniki, Greece Abstract Many old paintings suffer from the

More information

Feature Vector Similarity Based on Local Structure

Feature Vector Similarity Based on Local Structure Feature Vector Similarity Based on Local Structure Evgeniya Balmachnova, Luc Florack, and Bart ter Haar Romeny Eindhoven University of Technology, P.O. Box 53, 5600 MB Eindhoven, The Netherlands {E.Balmachnova,L.M.J.Florack,B.M.terHaarRomeny}@tue.nl

More information

Recipes for the Linear Analysis of EEG and applications

Recipes for the Linear Analysis of EEG and applications Recipes for the Linear Analysis of EEG and applications Paul Sajda Department of Biomedical Engineering Columbia University Can we read the brain non-invasively and in real-time? decoder 1001110 if YES

More information

CCA BASED ALGORITHMS FOR BLIND EQUALIZATION OF FIR MIMO SYSTEMS

CCA BASED ALGORITHMS FOR BLIND EQUALIZATION OF FIR MIMO SYSTEMS CCA BASED ALGORITHMS FOR BLID EQUALIZATIO OF FIR MIMO SYSTEMS Javier Vía and Ignacio Santamaría Dept of Communications Engineering University of Cantabria 395 Santander, Cantabria, Spain E-mail: {jvia,nacho}@gtasdicomunicanes

More information

D/A-Converters. Jian-Jia Chen (slides are based on Peter Marwedel) Informatik 12 TU Dortmund Germany

D/A-Converters. Jian-Jia Chen (slides are based on Peter Marwedel) Informatik 12 TU Dortmund Germany 12 D/A-Converters Jian-Jia Chen (slides are based on Peter Marwedel) Informatik 12 Germany Springer, 2010 2014 年 11 月 12 日 These slides use Microsoft clip arts. Microsoft copyright restrictions apply.

More information

IN the last decade, the field of digital image forensics has

IN the last decade, the field of digital image forensics has IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, VOL. XX, NO. XX, XXX XXXX Statistical detection of JPEG traces in digital images in uncompressed formats Cecilia Pasquini, Giulia Boato, Member,

More information

Here represents the impulse (or delta) function. is an diagonal matrix of intensities, and is an diagonal matrix of intensities.

Here represents the impulse (or delta) function. is an diagonal matrix of intensities, and is an diagonal matrix of intensities. 19 KALMAN FILTER 19.1 Introduction In the previous section, we derived the linear quadratic regulator as an optimal solution for the fullstate feedback control problem. The inherent assumption was that

More information

PRINCIPAL COMPONENT ANALYSIS

PRINCIPAL COMPONENT ANALYSIS PRINCIPAL COMPONENT ANALYSIS 1 INTRODUCTION One of the main problems inherent in statistics with more than two variables is the issue of visualising or interpreting data. Fortunately, quite often the problem

More information

Covariance and Correlation Matrix

Covariance and Correlation Matrix Covariance and Correlation Matrix Given sample {x n } N 1, where x Rd, x n = x 1n x 2n. x dn sample mean x = 1 N N n=1 x n, and entries of sample mean are x i = 1 N N n=1 x in sample covariance matrix

More information

IMAGE COMPRESSION-II. Week IX. 03/6/2003 Image Compression-II 1

IMAGE COMPRESSION-II. Week IX. 03/6/2003 Image Compression-II 1 IMAGE COMPRESSION-II Week IX 3/6/23 Image Compression-II 1 IMAGE COMPRESSION Data redundancy Self-information and Entropy Error-free and lossy compression Huffman coding Predictive coding Transform coding

More information

False Data Injection Attacks in Control Systems

False Data Injection Attacks in Control Systems False Data Injection Attacks in Control Systems Yilin Mo, Bruno Sinopoli Department of Electrical and Computer Engineering, Carnegie Mellon University First Workshop on Secure Control Systems Bruno Sinopoli

More information

Multi-scale Geometric Summaries for Similarity-based Upstream S

Multi-scale Geometric Summaries for Similarity-based Upstream S Multi-scale Geometric Summaries for Similarity-based Upstream Sensor Fusion Duke University, ECE / Math 3/6/2019 Overall Goals / Design Choices Leverage multiple, heterogeneous modalities in identification

More information

CS 231A Section 1: Linear Algebra & Probability Review

CS 231A Section 1: Linear Algebra & Probability Review CS 231A Section 1: Linear Algebra & Probability Review 1 Topics Support Vector Machines Boosting Viola-Jones face detector Linear Algebra Review Notation Operations & Properties Matrix Calculus Probability

More information

CS 231A Section 1: Linear Algebra & Probability Review. Kevin Tang

CS 231A Section 1: Linear Algebra & Probability Review. Kevin Tang CS 231A Section 1: Linear Algebra & Probability Review Kevin Tang Kevin Tang Section 1-1 9/30/2011 Topics Support Vector Machines Boosting Viola Jones face detector Linear Algebra Review Notation Operations

More information

MXPUF: Secure PUF Design against State-of-the-art Modeling Attacks

MXPUF: Secure PUF Design against State-of-the-art Modeling Attacks MXPUF: Secure PUF Design against State-of-the-art Modeling Attacks Phuong Ha Nguyen 1, Durga Prasad Sahoo 2, Chenglu Jin 1, Kaleel Mahmood 1, and Marten van Dijk 1 1 University of Connecticut, USA, 2 Robert

More information

Image Data Compression

Image Data Compression Image Data Compression Image data compression is important for - image archiving e.g. satellite data - image transmission e.g. web data - multimedia applications e.g. desk-top editing Image data compression

More information

Principal Component Analysis!! Lecture 11!

Principal Component Analysis!! Lecture 11! Principal Component Analysis Lecture 11 1 Eigenvectors and Eigenvalues g Consider this problem of spreading butter on a bread slice 2 Eigenvectors and Eigenvalues g Consider this problem of stretching

More information

Design and Development of a Smartphone Based Visible Spectrophotometer for Analytical Applications

Design and Development of a Smartphone Based Visible Spectrophotometer for Analytical Applications Design and Development of a Smartphone Based Visible Spectrophotometer for Analytical Applications Bedanta Kr. Deka, D. Thakuria, H. Bora and S. Banerjee # Department of Physicis, B. Borooah College, Ulubari,

More information

LEARNING from multiple feature sets, which is also

LEARNING from multiple feature sets, which is also Multi-view Uncorrelated Linear Discriminant Analysis with Applications to Handwritten Digit Recognition Mo Yang and Shiliang Sun Abstract Learning from multiple feature sets which is also called multi-view

More information

Estimation, Detection, and Identification CMU 18752

Estimation, Detection, and Identification CMU 18752 Estimation, Detection, and Identification CMU 18752 Graduate Course on the CMU/Portugal ECE PhD Program Spring 2008/2009 Instructor: Prof. Paulo Jorge Oliveira pjcro @ isr.ist.utl.pt Phone: +351 21 8418053

More information

Principal component analysis

Principal component analysis Principal component analysis Angela Montanari 1 Introduction Principal component analysis (PCA) is one of the most popular multivariate statistical methods. It was first introduced by Pearson (1901) and

More information

Reduced-Error Constant Correction Truncated Multiplier

Reduced-Error Constant Correction Truncated Multiplier This article has been accepted and published on J-STAGE in advance of copyediting. Content is final as presented. IEICE Electronics Express, Vol.*, No.*, 1 8 Reduced-Error Constant Correction Truncated

More information

MACHINE LEARNING. Methods for feature extraction and reduction of dimensionality: Probabilistic PCA and kernel PCA

MACHINE LEARNING. Methods for feature extraction and reduction of dimensionality: Probabilistic PCA and kernel PCA 1 MACHINE LEARNING Methods for feature extraction and reduction of dimensionality: Probabilistic PCA and kernel PCA 2 Practicals Next Week Next Week, Practical Session on Computer Takes Place in Room GR

More information