Image and Pattern Recognition


Person in charge: 
Johan Debayle

Email: debayle@emse.fr

Office: C3-03


General Description

Image and Pattern Recognition are mature but exciting and fast developing fields, which underpin developments in cognate fields such as computer vision, image processing, text and document analysis and neural networks. It is closely akin to machine learning, and also finds applications in fast emerging areas such as biometrics, bioinformatics, multimedia data analysis and most recently data science.

 

The objective of this GP is to know the necessary mathematical and computational tools and master the concepts of geometrical characterization of shapes (signals, images, point pattern) to get basic knowledge about machine learning on images for real applications.

 

At the end of this toolbox, the student will be able to manipulate the main aspects of modern geometry, use concepts and results to solve concrete problems, such as the extraction of geometrical, morphometrical and textural information in image analysis as well as the characterization, modeling and simulation of point patterns or spatial object distributions. He will also be able to use some basic machine learning techniques to answer image and pattern recognition applicative problems, such as the automatic detection of cancerous skin lesions in biomedical imaging.

 

Important note: This toolbox will be used to validate a part of the Master of Science "Mathematical Imaging and Spatial Pattern Analysis" to which students can enroll in the third year of the ICM cycle to obtain a double degree. For more information, contact Johan DEBAYLE (debayle@emse.fr)

 

Content

This GP of 80h consists of three UPs:

  • Mathematical Geometry (33h)
  • Computational Geometry (24h)
  • Applications (23h)


Keywords

Convex geometry, integral geometry, fractal geometry, stochastic geometry, image analysis, point patterns, convex hull, Delaunay triangulation, alpha-shapes, statistical shape analysis, machine learning, neural networks


Consitence between UPs

The first UP gives the mathematical tools for the extraction of geometrical, morphometrical and textural information from images. The second UP gives the computational tools to characterize, model and simulate point patterns or spatial objects distributions. The third UP is focused on one or two real applications of image and pattern recognition by using the previous mathematical and computational tools.


Consitence with other GPs

This GP is strongly consistent with the TB « Introduction to Image Processing », as well as with other pedagogic units around the digital sciences: TB High Performance Calculation, MAJ Data Science, TB Fundamentals of EF Calculations, DEFI Big Data...


Person in charge

Yann Gavet / gavet@emse.fr / office C3-22


Objective

The objective of this UP is to know the concepts and classical tools of computational geometry to characterize, model and simulate point patterns or spatial objects distributions.


Content

This UP is mainly composed of lectures (10,5h), tutorials using Matlab (10,5h).


Evaluation

Two exams, theoretical (1,5h) and practical using Matlab (1,5h), will evaluate the learning of this UP.


Keywords

Point patterns, convex hull, Delaunay triangulation, alpha-shapes, statistical shape analysis

Person in charge

Johan Debayle / debayle@emse.fr / office C3-03 


Objective

The objective of this UP is to master the mathematical concepts and tools for the extraction of geometrical and morphometrical characteristics from images for using them in learning tools.


Content

This UP is mainly composed of lectures (15h), tutorials using Matlab (15h).


Evaluation

Two exams, theoretical (1,5h) and practical using Matlab (1,5h), will evaluate the learning of this UP.


Keywords

Convex geometry, integral geometry, fractal geometry, stochastic geometry, image analysis

Person in charge

Johan Debayle / debayle@emse.fr / office C3-03


Objective

The objective of this UP is to get basic knowledge about machine learning on images for real applications. The training will be done on one or two real applications of image and pattern recognition, in the form of a case study, by using the mathematical and computational tools provided in the previous UPs.


Content

This UP is mainly composed of tutorials (in the form of a case study) using Matlab (23h).


Evaluation

A report on the selected subject will be asked to the student to evaluate his ability to use some basic machine learning techniques to answer image and pattern recognition applicative problems. 


Keywords

Machine learning, neural networks, interpretation, decision