E9 Signal Processing, Acoustics and Bioengineering

 


 

E9 201 (AUG) 3:0

Digital Signal Processing

 

Discrete-time signals and systems, frequency response, group delay, Z-transform, Convolution discrete fourier transforms (DFT), fast fourier transform (FFT) algorithms, discrete Cosine transforms (DCT),  Discrete Sine transforms (DST), relationship between DFT, DCT and DST; Design of FIR and IIR filters, finite word length effects, Hilbert transforms, Hilbert transform relations for Causal signals, KarhunenLoeve transforms, introduction to linear prediction, bandpass sampling theorem, Bandpass signal representations.

 

D Narayana Dutt and Chandra Sekhar S

Proakis and Manolakis,  Digital Signal Processing,  PHI.
AV Oppenheim and RW Schafer, Discrete-time Signal Processing. Prentince Hall 1998

 

E9 202 (AUG) 3:0

Advanced Digital Signal Processing

 

Non-linear signal processing; non-linear filters. Non-Gaussian models, generalized Gaussian and stable distributions, robust estimation.  Medium smoothers, rank-order filters, weighted median smoother. Introduction to order statistics, joint densities, momemts.  Weighted medium filtering, link between linear and non-linear smoothers and filters, Mallows Theorem.  Generalized median filtering: L-estimator, tefilter, optimality  Zero-crossing (ZC) based spectral analysis, dominant frequency principle. ZC and level-crossing based signal decomposition. Auditory models.

 

T V Sreenivas

 

Pre-requisite: E9-201 or equivalent.

 

Arce, G.R.,  Non-linear signal processing: A statistical approach, Wiley, 2004.

Astola, J., and Kuosmanen, P.,  Fundamentals of non-linear digital filtering, CRC Press, 1997.

Kedem, B., Time series analysis by higher order crossings, IEEE Press, 1994.

Maravasti, F., A unified approach to zero-crossings and non-uniform sampling, 1990.

 

E9 212 (JAN)  3:0

Spectrum Analysis

 

Estimation of PSD from finite data:  Nonparametric methods: Periodogram,  Properties, Bias and Variance analysis – BT method, Window design considerations, Time-Bandwidth Product and Resolution, variance tradeoffs in window design. Modified Periodogram methods: Bartlett, Welch. Parametric methods for Rational Spectra: Covariance properties of AR, MA, ARMA processes, Yule-Walker method, Levinson-Durbin algorithm. Multivariate ARMA  processes: State-space representation, subspace parameter estimation.  Parametric methods for line spectra (sinusoids in noise), models of sinusoids in noise, higher-order YW method, Pisarenko, MUSIC, ESPRIT, Prony methods.  Formulation of the spatial spectrum problem.  Introduction to Higher-Order Spectra.

Matlab based assignments.

 

K V S Hari

 

Stoica and Moses, Introduction to Spectral Analysis, PH 1997.

 

E9 213(JAN) 3:0

Time-Frequency Analysis

 

Time-Frequency distributions: Time and frequency description of signals, instantaneous frequency, the analytic signal, the uncertainty principle, densities and characteristic functions, global and local averages, the short-time fourier transform (STFT), filterbank interpretation of STFT, the Wigner distribution , general approach and the kernel method, bilinear time-frequency distributions, Wigner’s theorem, multi component signals, instantaneous band width, positive distributions satisfying the marginals, gabor transform Spaces and bases: Hilbert space, Banach space, orthogonal bases, orthonormal bases, Riesez bases, biorthogonal bases, shift-invariant spaces, Shannon sampling theorem, frames, b-splines.

Wavelets: Wavelet Transform, Real Wavelets, Analytic Wavelets Dyadic Wavelets transforms, wavelet bases, multi resolution analysis, Two-Scale equations, conjugage Mirror filters, vanishing moments, regularity, Lipschitz regularity, Strang-Fix conditions, The Notion of Compact support, Shannon, Meyer, Haar and Battle-Lemarie Wavelets, Daubechies Wavelets, Relationship between Wavelets and Filterbands.

 

Chandra Sekhar S

 

L Cohen, Time Frequency Analysis Prentice Hall 1995

S Mallat, A Wavelet Tour of Signal Processing, The Sparse Way, Elsevier, Third Edition, 2009

 

E9 231 (AUG) 3:0

Digital Array Signal Processing

 

Wave Fields, underlying wave equations, scalar and vector fields, spectral  representation, propagation in open and confined media.   Sensor array systems: Linear equispaced, circular, planar, random arrays. Direction of arrival estimation: source waveform estimation. Beam forming, subspace methods (MUSIC, ESPRIT), spatial smoothing, performance analysis.  Applications to acoustic source separation and wireless communication.

 

K V S Hari

 

Johnson and Dudgeon, Array Signal Processing Concepts and techniques, Prentice Hall, 1993.

Naidu, P.S., Sensor Array Signal Processing, CRC Press, 1999.

Lecture notes and current literature.

 

E9 241 (AUG) 2:1

Digital Image Processing

 

Representation of two dimensional signals, sampling, quantization and reconstruction. Digital images, human visual perception. Transforms: DFT, DCT, KLT, wavelet, filtering, edge detection, image restoration, compression, segmentation, applications.

 

K Rajgopal

 

Anil K Jain, Fundamentals of Digital Image Processing, Prentice Hall, 1989.

Gonzales, R.C., and Woods, R.E., Digital Image Processing, Addison Wesley, 1993.

 

 

E9 242 (JAN) 3:0

Selected Topics in Image Processing

 

Image segmentation and clustering: mean-shift, graph cut. Image pyramids and texture analysis. Linear/non l linear scale space theory: Scale Invariant Feature Transform (SIFT). Visual tracking: mean-shift, particle filters, feature-based. Background modeling, surveillance and monitoring: event detection/recognition. Face detection and recognition, motion analysis and segmentation, graphical models, Markov random fields and applications. Basics of 3-D vision.

 

K R Ramakrishnan

 

Forsyth, A.A., and Ponce, J., Computer Vision: A Modern Approach, Pearson Education, 2003

Paragios, N., Chen, Y., and O. Faugeras, O. (Eds), Handbook of Mathematical Models in Computer Vision, Springer, 2006.

Bishop, C.M., Pattern Recognition and Machine Learing, Springer, 2006.

Current Literature

 

 

E9 243 (JAN) 3:0

Computer Aided Tomographic Imaging

 

Introduction to principles of tomography and applications, tomographic imaging. Radon transform and its properties, mathematical framework. Introduction to X-ray tomography, emission computer tomography, magnetic resonance imaging systems. Projection and Fourier slice theorem. Scanning geometries: translate and rotate, translate-rotate, rotate on a circular trajectory for 2-D imaging and helical or spiral scan trajectory for 3-D imaging. Transform domain algorithms: Fourier inversion algorithms, filtered back projection algorithms – reconstruction with non-diffracting sources, parallel projections and fan projections for 2-D and cone beam projections on circular and spiral trajectory for 3-D reconstruction. Computer implementation, iterative reconstruction techniques: algebraic reconstruction techniques, statistical modeling of generation, transmission and detection processes in X-Ray CT, artifacts and noise in CT images. Image reconstruction with incomplete and noisy data, applications of Radon transform in 2-D Signal and Image processing.

 

K Rajgopal

 

Kak, A.C., and Slaney, M., Principles of Computerized Tomographic Imaging, IEEE Press, 1988.

Herman, G.T., Image Reconstruction from Projections, Implementation and Applications: Topics in Applied Physics, Vol 32, Springer Verlag, 1979.

Natterer, F., The Mathematics of Computerized Tomography, SIAM Classics In Applied Mathematics, Vol. 32, 2001.

Natterer, F., and Wubbeling, F., Mathematical Tools in Image Reconstruction, SIAM, 2001.

 

 

E9 245 (AUG) 3:1

Selected Topics in Computer Vision

 

This course will develop the use of multiview geometry in computer vision. A theoretical basis and estimation principles for multiview geometry, dense stereo estimation and three-dimensional shape registration will be developed.The Use of these ideas for building real-world solutions will be emphasised.  Topics Stereo estimation: current methods in depth estimation * 3D registration : ICP and other approaches Multiple view geometry: projective geometry, multilinear relationships in images, estimation.

 

Venu Madhav Govindu

 

Pre-requisites: Computer Vision (E1.216) or permission of the instructor.

 

Hartley, R., and Zisserman, A., Multiple View Geometry in Computer Vision, Second Edn, Cambridge University Press, 2004.

Faugeras, O., and Luong, Q-T.,The Geometry of Multiple Images, MIT Press 2001.

Current literature

 

E9 261 (JAN) 3:0

Speech Information Processing

 

Human speech communication: speech production, perception, phonetics. Time-varying signal analysis: short-time Fourier transform, spectrogram, quasi-stationary analysis: cepstrum, linear-prediction models; line spectral pair, Mel frequency cepstral coefficients sinusoidal models, Principles of Speech synthesis, pitch and time scale modification. VQ and Gaussian mixture model. Speaker recognition.

 

 A G Ramakrishnan

 

Pre-requisite: E9-201

 

Quatieri, T.F., Discrete-time speech signal processing, Prentice-Hall, 2002.

Rabiner, L.R., and Schafer, R.W., Digital Speech Processing, Prentice-Hall

Douglas O’shoughnessy, Speech Communication, IEEE Press 2000

Taylor, P., Text-to-Speech Synthesis, Cambridge Univ. Press, 2009.
Gold, B., and Morgan, N., Speech and Audio Signal Processing, John Wiley, 2000.


   
E9 262 (JAN) 3:0

Automatic Speech Recognition Algorithms

 (Stochastic models for speech/audio)

 

Human speech communication, concept=> signal=> concept & levels of information.  Discrete and continuous representations, signal representation as a pattern; structure representation through lexicon, grammar. ASR: text recognition, speaker recognition, language identification, keyword spotting. Gaussian models and Bayesian inference; maximum likelihood parameter estimation. Mixture Gaussian models, EM algorithm derivation; relation to K-means algorithm, LBG algorithm and EM generalization. Application to speaker-ID.  Units of speech: linguistic, acoustic and stochastic; segmentation problem. Dynamic programming introduction. maximum-likelihood segmentation; segment clustering and automatic sub-word units.  Graphical models and Markov models; Language modeling, N-grams and their estimation.  Tree structured language model, minimum entropy decision tree algorithm; language perplexity measure. Application to spoken language-ID.  Hidden Markov model (HMM): Markov structure for latent variables; Gaussian density, discrete density, mixture Gaussian and semi-continuous density models. HMM evaluation, training and decoding problems: forward-backward algorithm, Baum-Welch algorithm, Viterbi algorithm, segmental K-means (SKM) algorithm. HMM duration density and explicit duration modeling and modified EM algorithm.  Finite state network (FSN) of HMMs and lexicon building. Continuous speech recognition (CSR) through FSN decoding using time-synchronous Viterbi algorithm. Viterbi beam search for large vocabulary CSR. Speaker adaptation: maximum likelihood linear regression (MLLR) and eigen-voice approaches. Not checked. Unable to understand.

 

T. V. Sreenivas

 

Pre-requisite: E2-202 Random Processes or equivalent.

 

Huang, X., Acero, A., and Hon, H., Spoken Language Processing, Prentice Hall, 2001.

Bishop, C.M., Pattern Recognition and Machine laearning, Springer, 2006.

Rabiner, L.R., and Juang, B.H., Fundamentals of Speech Recognition, Prentice Hall, 1993.

 

 

E9 271 (JAN) 3:0

Space-Time Signal Processing and Coding

 

Brief review of single-input single-output (SISO) communication systems.  Performance of SISO systems in fading channels.  Motivation for Space-Time (or Multiple-Input Multiple-Output (MIMO) communication systems.  Capacity of MIMO systems.  Space-Time codes: Space-Time Trellis codes and Space-Time Block codes; Design Criteria for code constructions; Constructions using Orthogonal designs and their variations; Algebraic techniques for space-time codes; Decoding algorithms for Space-Time Codes.   Distributed Space Time Coding.

 

B Sundar Rajan

 

Pre requisites: Digital Communication

 

Current literature

 

E9 281 (JAN) 3:0

Biomedical Signal Processing

 

EEG and ECG signals: Genesis, monitoring, measurement and uses. Time and frequency domain analysis of signals: Morphological studies, correlation, spectral analysis. Linear prediction technique: AR model and its implementation, inverse filter. Homomorphic processing: generalized superposition, complex cepstrum, minimum phase component. Nonlinear dynamics and chaos: fractal dimension, correlation dimension, Lyapunov exponent. Artifacts: Types, detection and minimization. Applications to biomedical signals.

 

D Narayana Dutt

 

Tompkins, W.J. (ed.), Biomedical Signal Processing, Prentice Hall, 1993.

Rangayyam, R.M., Biomedical Signal Analysis – A Case-Study Approach, John Wiley, 2002.

Chellis, R.E., and Kitney, R.I., Biomedical Signal Processing, in IV parts, Medical and Biological Engg. and Current Computing, 1990-91.

Current Published literature.

 

 

E9 291 (AUG) 2:1

DSP System Design

 

Need for special digital signal processors. DSP architecture: Von Newmann vs Harvard architecture, architectures of superscalar and VLIW fixed and floating point processors, on chip components: timers, boot ROM, RAM, CODECS, McBSP, DSP bias. Design aspects: memory, host port interface, CODECS and RTDX. DSP development tools: assembler, C-compiler, linker and loader. DSP algorithms: FIR/IIR filters, FFT, adaptive filters, sampling rate converters and DCT. DSP applications: FIR and FFT with RTDX and MATLAB, voice scrambler, multirate filter. Weekly laboratory exercises involving DSP starter kits, interfacing, coding, and use of software tools.

 

G N Rathna

 

Rulph Chassaing, Digital signal processing and applications with the C6713 and C6416 DSK, Wiley, 2005.

User’s manuals of various fixed and floating point DSPs.

Application Guides from DSP manufacturers.

Current literature.

 

 

E9 292 (JAN) 2:1

Real-Time Signal Processing with DSP

 

Implementation of discrete-time systems, DSP device architecture and programming (TMS320C6x), FIR/IIR digital filter design, multirate DSP, power spectrum estimation, linear prediction and adaptive filtering, real-time system development, DSP programming, code composer studio and DSP BIOS, spawning and controlling tasks and data I/O, real-time scheduling analysis, load analysis, queues, semaphores and mailboxes, real-time data exchange using Lab View.

Mini project.

 

K Rajgopal and G N Rathna

 

Pre-requisite: Knowledge of Digital Signal Processing

 

Nasser Kehtarnawaz, Real–Time Digital Signal Processing based on TMS320C6000, Elsevier, 2004.

TMS320C6x Data Sheets from TI