Introduction to Digital Signal Processing (Discrete-time Signal Processing) Prof. Ja-Ling Wu Dept. CSIE & GINM National Taiwan University Overview Introduction to DSP Information Theory and Coding Tech. C&C Digital Speech Processing Multi-D DSP Digital Image Processing Multi-rate DSP VLSI DSP Advanced topics in DSP Data Compression Error Control coding Recognition Understanding DSP-processor Application-specific system efficient communication system Digital Modulation Information Security Reliable communication system Multimedia Security Secure communication system Introduction DSP theory has been applied to a variety of problems such as: •Biomedical data processing : EEG, ECG(EKG), CT… •Digital Audio : CE-Disk, Audio-CD •Sonar and Radar processing : homeland security, video surveillance, military applications •Speech processing : recognition •Data Communication : ISDN, Digital Communications •Reliable data storage of computerized information : digital storage with error correction capability •Seismic Signal Processing : oil exploration, underwater mapping •Image Processing and Vision : Data Compression, Computer Vision •Error Control Coding : Reliable communication •Information Security : steganograph / data hiding / watermarking / forensics Signal Classification Continuous time Discrete time x(t) x(t) Continuous amplitude Analog t x(t) Discrete t Digital t x(t) Discrete amplitude Sample-data t Block Diagram of digital processing for analog waveforms Band-limited signal Analog Signal input Low-pass filter Analog / Digital Conversion •Sampling •Quantization DSP double- period sequence Analog Signal output Low-pass filter digital Digital Processes Digital / Analog Conversion •Sample-and-hold digital Digital Devices (VLSI) Digital Computers DSP • Advantages of digital processing – Reliable – easy to be stored and/or transmitted – Flexible – Accurate – Faster • Disadvantage: Easy to make • Challenges ‒ Exact copy!! – Optical processing ‒ Good forgery!! – Bio-computing Serious IPR threats What does “digital” mean ? Analog Waveform x(t) △z Quantization step Discrete in amplitude t △t : Sampling period Discrete in time Remarks 1. What is “DSP” ? keep what we want and eliminate what don’t as ________ as possible ! f(t) trend prediction – Much – Precisely ? – Soon 2. Δt (Sampling period) Sampling Theorem t T T+n – Fourier Analysis / Transform video motion estimation – Interpolation / Extrapolation 3. Δz (Quantization step) Finite-Wordlength Effects – Available Hardware Support – Precision Requirement uniform vs. non-uniform scalar vs. vector quantization 4. What kind of signals can we really process ? – Bandlimited – finite-dynamic-range – If the signal is stochastic, some statistic properties must be known; say, mean, variance, acf, psd, …etc. (random/stochastic process) – long-duration signal (such as: voice/speech signal) short-time analysis. (sliding window) Course outline 1. Introduction 2. Signals and Systems • • • • • • 3. Z-transform System function LTI-system digital convolution Sampling Theorem System stability Fourier Response of a System • • • • Fourier Transform DFT Fast Fourier Transform Convolution Theorem 4. Digital Filters • FIR filter • IIR filter 5. Quantization and Finite Wordlength Effects 6. Specific topics in DSP References 1. Digital Signal Processing by Roberts & Mullis (Addison Wesley) 2. Discrete time Signal Processing by Oppenheim (Prentice-Hall) 3. Signal Processing First by James H. McClellan, Ronald W. Schafer, Mark A. Yoder Course Information • Lecturer: Ja-Ling Wu (wjl@cmlab.csie.edu.tw) • TA: Yin-Tzu Lin (known@cmlab.cie.ntu.edu.tw) Yun-Chung Shen (cazindo@cmlab.csie.ntu.edu.tw) • Lecture Notes – http://www.cmlab.csie.ntu.edu.tw/~dsp/dsp2010 • Grades – – – – 40% Homework 15% Quiz 1 : in-class + take home(extra 15% of Quiz 1) 15% Quiz 2 : in-class + take home(extra 15% of Quiz 2) 30% Final • Writtten Quiz (60% of final) • Monograph (40% of final) • MATLAB will be used in homework