Wednesday, 23 August 2017

INFO-H-501

From Image

Jump to: navigation, search

INFO-H-501 Pattern recognition and image analysis

This course is the fusion of two pre-existing courses on Image analysis (Pr O. Debeir) and the MATH-H-500 course of Pr Ch. Decaestecker.

Lien vers la Fiche résumée de cours ULB [1]

Télécharger l'horaire .iCal

Visualiser la page Gehol Q1 Gehol Q2l

Lectures content



PATTERN RECOGNITION (Part I)

  • Introduction
    • Basic principles
    • Bayesian approach of supervised classification
    • Examples of method
  • Decision / classification trees
    • Introduction
    • Induction Methods
    • Benefits - Limitations - Extensions
    • Biomedical Applications
  • Artificial neural networks
    • History - Introduction
    • Basic principles
    • Components and Architecture
    • Supervised learning algorithms and a priori probability managment
    • Alternative: Support Vector Machine (SVM)
    • Unsupervised Networks
    • Practical aspects
  • Evaluation and comparison of performance
    • Evaluation of a supervised classification model
    • Model comparison and selection

Slides

slides

former french version



IMAGE ANALYSIS (Part II)

  • segmentation
    • intro
    • optimal, entropy, percentile threshold
    • Otsu threshold
    • 2D threshold
    • gaussian mixture
    • condensation algorithm
    • graph-cuts
    • hierarchical
    • region growing
    • split and merge
    • image statistics (for segmentation)
  • model based segmentation
    • live wire &
    • live wire & Dijkstra’s algorithm
    • example of active contour applications
    • snakes
    • level sets
    • edge-less active contours
    • Active Shape models
    • Active Appearance models
  • shape description
    • chain coding
    • polygonal approximation
    • spectral description (intensity)
    • 2D shape descriptors (contours)
    • 2D shape descriptors (region) area, euler, elong,...
    • moments
    • Minkowski fractal dimension
  • texture analysis
    • fractal analysis: Hurst coeficient
    • spectral approach
    • Gabor filter
  • detectors
    • Harris corner detection
    • Fast
    • sift, surf, others
    • DoG pyramid
    • Robust object detection
    • Hough transform
  • tracking
    • optical flow
    • meanshift 2D/3D
    • cell tracking
    • particle filter
    • body skeleton fitting on range data (particle filter)
    • face tracking using color histogram
    • 3D reconstruction from n-range images
  • object recognition
    • the general recognition problem
    • face detection
    • finger print recognition
    • microscopy image analysis
    • traffic analysis, one class classifier
    • bag of visual word
    • remote sensing land use classification

Notebooks

A new version of the course material is now based on ipython notebooks Info-h501 on github (PART II)

Labs on GitHub

online view of the notebooks (version 2016-2017)

in case of problem, a snapshot is available here, the version may be outdated, the very last version is on the github repository.

Version control with Git

Download data for Lab 4

Slides and code examples (former version)

slides

supporting material

Laboratories

online view of the labs (version 2015-2016)

in case of problem, a snapshot copy is available here, the version may be outdated, the very last version is on the github repository.


Info-H-501 Labs 1 - 6

Links

reference books used

Annotated image processing bibliography

List of publications in computer graphics conferences (Siggraph, Siggraph Asia, EuroGraphics, HPG, I3D, ...)

Feature Extraction & Image Processing for Computer Vision (Third Edition)