The course is split into two modules.

The first module introduces the fundamental tools for digital signal processing. It mainly focuses on signal analysis and filtering concerning audio and video sources but the theoretical part allows the student to easily extend his/her knowledge to other digital signal processing fields. The course aim at providing the participants with basic tools for digital signal filtering and analysis, and with the knowledge to design specific processing tools for 1D and 2D discrete signals.

The goal of the second module is to provide students with advanced knowledge and skills on multimedia signal processing applications, with a specific focus on coding of multimedia objects. This course covers relevant applications of digital signal processing with specific reference to multimedia communication, e.g. audio, image and video coding. Furthermore, the course gives an insight on widely-adopted international coding standards such as MPEG Audio, JPEG and MPEG Video, among others.

Both modules focus on theoretical and practical aspects with laboratory classrooms in MATLAB where students can directly apply and verify learnt notions.

Module I: Fundamentals of digital signal processing

• Review of Analog Signal Processing theory:
  • Some fundamentals of Analog Signal Processing will be recalled (sampling, quantization, Fourier Transform and Series, noise description...).
• Introduction to Discrete Signals Transforms:
  • The z-transform and the Discrete Fourier Transform will be described with their properties and application to filter design and analysis.
• Introduction to Digital Filters:
  • This course part will cover Time-domain and space domain filter representations, transfer function analysis, frequency response analysis, Finite and Infinite impulse response implementation; stability analysis
• Windowing and Short Time Fourier Transform:
  • Overview of windows and real-time processing; overlap and add, overlap and save,Short Time Fourier Transform for real time processing
• Introduction to Multirate Processing:
  • Downsampling, upsampling, polyphase filters and perfect reconstruction filter banks will be introduced.
• Multimedia applications:
  • Practical filter design for audio signals, basic image processing (e.g., edge extraction, histogram processing,rescaling).

Module II: Fundamentals of coding

• Source coding:
  • Discrete memoryless sources, discrete sources with memory, entropy of a source, uniquely decodable and prefix codes, Shannon’s source coding theorem, Huffman coding, arithmetic coding, run length coding.
• Quantization:
  • Uniform scalar quantization, Lloyd-Max scalar quantization, entropy constrained scalar quantization, rate-distortion theory, vector quantization.
• Predictive coding:
  • Linear predictive coding, DPCM, delta modulation, predictive coding gain.
• Transform coding:
  • Linear transforms, unitary transforms, linear approximation, non-linear approximation, KLT, DCT, transform coding gain, bit allocation, sub-band coding, wavelet transform, 2D transforms.
• A review of Waveform Coding of audio signals:
  • PCM, DPCM, Delta Modulation, ADPCM. Lossless compression techniques.
• Speech coding:
  • Vocal tract modeling, LPC, pitch extraction, voiced/unvoiced detection, analysis by synthesis.
• Audio coding:
  • Fundamentals of psychoacoustics, frequency masking, temporal masking, filter banks (PQMF, MDCT), bit allocation and entropy coding. Coding standards: MPEG-Audio, Advanced Audio Coding (AAC), AC3.
• Image coding:
  • Human visual system, visual redundancy and irrelevancy, lossless and lossy image coding, transform coding and quantization. Coding standards: JPEG.
• Video coding:
  • DPCM, motion estimation, coding of prediction residuals, coding of motion vectors, rate-distortion optimization. Coding standards: MPEG-x, H.264/AVC.
• Other multimedia applications:
  • Multimodal analysis, image intensity pointwise operations, color handling, histogram equalization, morphological processing, feature analysis, matching, registration, forensics.

Innovative Didactics

The "Flipped Classroom" methodology will be applied to practical aspects concerning better understanding of Matlab techniques for a proper signal processing.

Students are required to know the basic principles of signal processing and to have basic programming skills.

Module I: Fundamentals of digital signal processing

The assessment will be based on a written exam at the end of the course. The written exam consists of numerical exercises, theoretical questions and programming exercises (in MatLab®); no further oral exams or projects are considered. Further details on the exam procedure are available on the lecturer’s homepage.

Module II: Fundamentals of coding

The assessment will be based on a written exam at the end of the course. The written exam consists of two parts. The first part contains numerical exercises and theoretical questions. This will assign up to 23 points. The second part consists in a programming test. This will assign up to 10 points. 30 cum laude will be assigned when the total score is higher than 31.

Concerning the assessment of the whole exam (Module I and Module II)

In order to pass the whole Multimedia Signal Processing exam, a positive grade must be obtained in both modules and the final mark will be the average of the two marks rounded up the next integer.

In evaluating the average between the two modules a 30 cum laude mark in a module will be considered as 30; in order to get 30 cum laude as a final mark, the mark of at least one module must be 30 cum laude.

In every exam date one module will take place in the morning while the other in the afternoon (the sequence will be published in the web poliself) but the exams of both modules will take place in the same day; anyway you can choose to take on a single module or both modules in the same day. Once you get a positive mark in a module this mark will be automatically frozen until you take again the same module in a following session.

Once you will get a positive mark in both modules the final mark will be evaluated and then published for the publishing period on the web poliself ; at the end of this period it will be automatically recorded. If you do not want to record that final grade you have to refuse it from the poliself: in that case both marks will be restored in a "frozen" state in order to allow you to take on again a module (or both of them). Anyway if you have a frozen mark in both modules at the end of each exam to which you signed in, the final average mark will be automatically published and, after the publishing period, automatically recorded: so, if you want to improve your final mark you have to remember to refuse the published final grade.