Signal processing

Building on the concepts of signal processing covered in 1A, the aim here is to acquire skills in signal processing dedicated to speech and biomedical applications, as well as digital image and video processing.
Two projects, one dealing with augmented reality and the other with the analysis of audio, speech or biomedical signals, carried out in Matlab, enable students to master the implementation of algorithms.
Keywords:

Digital image,
Color,
Computer vision,
Augmented reality,
Geometric transform,
Segmentation,
Fourier transform,
Filtering.

The prerequisites for this module are the 1A signal processing module (module CO6SFSS0).

Part 1: Digital image and video processing (11h CM, 12h TD, 12hTP - Lecturers: Marc Donias + lecturer)
Chapter 1: Introduction

Contents
Applications

Chapter 2: Imaging chain

Imaging chain
Sources
Acquisition and restitution
Characteristics of the human visual system
Color spaces
Stereovision

Chapter 3: Fourier transform

Continuous domain
Discrete domain
Extension to the 2d case

Chapter 4: Processing

Point-to-point operators
Filtering
Edge detection
Geometric transformations
Transforms

Part 2: Signal processing (6h CM, 6h TD, 6hTP - Contributors: Pierrick Legrand + contributor)
Chapter 1: Theoretical background and applications specific to the course
Chapter 2: Frequency characterization of a discrete-time signal

Difference between Fourier transform of a discrete sequence and discrete Fourier transform
Fast Fourier Transform (FFT). Zero padding
Spectrogram-based time-frequency analysis

Chapter 3: Digital filtering
Influence of poles and zeros (stability, minimum-phase filter, impact on filter frequency response - resonances, total rejection, etc.).)
"Philosophy of TD/TP/Projects":
Part 1: Digital image and video processing

The aim is to complement and illustrate the course by using the Matlab tool to manipulate digital images for display, frequency analysis and elementary processing (chroma-keying, convolutional linear filtering, edge detection, enhancement). An augmented reality project, also implemented in Matlab, allows us to tackle different aspects (color segmentation, geometric transformations, video overlay) through a concrete, realistic application.

Part 2: Signal processing

The aim is to use Matlab to implement the concepts introduced in class, in particular FFT, digital filtering and windowing, in a project that could involve enhancing a speech signal, setting up functions dedicated to audio signal processing, as in the case of a mixing desk, or even techniques for characterizing biomedical signals.

Applied mathematics