Cours de traitement d'images et vision par ordinateur pour le Master 2 Signal Image Synthèse à l'Université Paris-Est Marne-la-Vallée de Janvier 2015 à Mars 2015


- Mercredi 28/01/15 de 14H à 16H cours 1
Image processing basics : sampling, quantification, aliasing, histograms, convolution, smoothing, canny edges detection Overview and references Slides
- Jeudi 29/01/15 de 14H à 16H tp 1
Enoncé TP1, correction
- Mercredi 04/02/15 de 14H à 16H cours 2
Continuous optimization applied to image processing : gradient descent, linear least squares, reweighted least squares, majoration minimization. Some applications: Rudin-Osher-Fatemi denoising, inpainting with total variation Slides
- Jeudi 05/02/15 de 14H à 16H tp 2 correction
- Mercredi 18/02/15 de 14H à 16H cours 3
Segmentation : how can we partition an image along edges into homogeneous regions ? kmean, graph-cuts, active contours Slides
- Jeudi 19/02/15 de 14H à 16H tp3
Enoncé TP3, correction
- Mercredi 25/02/15 de 14H à 16H cours 4
Detecting geometric primitives and rigid objects : Hough transform , RANSAC, iterative closest points (ICP) Slides
- Jeudi 26/02/15 de 14H à 16H tp 4 correction
- Mercredi 04/03/15 de 14H à 16H cours 5
Sparse and dense points correspondances between images : interest points, patch matching, optical flow Slides
- Jeudi 05/03/15 de 14H à 16H tp 5
Enoncé TP5, correction Projects
- Jeudi 19/03/15 de 14H à 18H Soutenances de projets


You will implement in python the method presented in one of the following papers and write a small report of about 4 pages where you will quickly present the main concept, present some implementation/optimization details that where missing from the original paper, present various result and discuss the choice of paramters. I recommand the use of LateX to write the report. You can use the great BaKoMa TeX editor that allows to see the generated document been updated in real time as well as direct editing in the generated document.

2014 Projects

A non-local algorithm for image denoising
GrabCut - Interactive Foreground Extraction using Iterated Graph Cuts
Real-Time Tracking of Non-Rigid Objects using Mean Shift
Stereo Matching and Graph Cuts
PatchMatch Stereo - Stereo Matching with Slanted Support Windows

2015 Projects

Fast Dichotomic Multiple Search Algorithm for Shortest Cirular Path
LSD a Line Segment Detector
Shape Context and Chamfer Matching in Cluttered Scenes
Star Shape Prior for Graph-Cut Image
The Generalized PatchMatch Correspondence Algorithm
Shape priors in variational image segmentation: Convexity, lipschitz continuity and globally optimal solutions