Hoppa till innehållet

Tillämpad multivariat bildanalys (MIA) och regression (MIR)

Publicerad: 15 november 2016

Course syllabus

Course name: Applied Multivariat image analysis  (MIA) and regression (MIR)

ECTS/HP: (7.5)

Course syllabus is decided by: Head of Department Elisabet Kassfeldt

Examiner: Professor Olle Hagman

Entry requirements: Basic knowledge in Image processing and Multivariate analysis

Intented learning outcomes: The purpose of this course, is to define, study and build knowledge on multivariate image analysis applied to different sensor technologies and inspected materials, mechanisms and real world observations.

The purpose is that you should get to learn and practice multivariate modeling and within the scope of subjective human observations and multivariate image based measurements. One key factor is the use of visualisation as a tool for analysis och complex measurements. The knowledge building strategies and supervision will be adapted to the PhD students ongoing projects, chosen process and needs.

After completion of the course you should be able to:

  • Identify, design and visualize the processes based on multivariate images needed for understanding a comples material or mechanism
  • Identify your needs of new knowledge and capability to find the image based technology and information needed for a identifying special process.
  • Use the imaging sensors and MIA/MIR technology for your research project
  • Present a project at a quality level that can be published in at an international conference.

Course content:

The course consists of three main modules:   

  • lectures and case studies, (1hp)

Day 1

Startup Introduction and examples

Digital images, image processing, 2D-sensors, Spatial and spectral resolution, multivariate images……

Day 2

Multivariate image processing, scaling, visualization, residuals and analyze.

Day 3

Examples and case studies: VIS, NIR, Satellite images…….

Laborations (2 hp)

  • I: MIA on imaging spectrometers NIR and VIS
    • Imaging spectrometer NIR
    • Imaging spectrometer VIS
    • Hard and Software for visualization
  • II: Multivariate image regression on Multisensor images.
    • Combination of images from different sensors.
    • Scaling
    • Residual images

Individual projects (4.5 hp)

Based on research question and need for each student a MIA project is defined and completed.

  • Research question defined
  • Choise of sensor system: VIS, NIR or other multi- or univariate imaging sensors system.
  • Experiment
  • MIA/MIR Analysis: Scoreplots residuals and predictions etc.

Course methods: Lectures Laborations and applied student projects

Examination format: Laboration reports, project report and presentation.

Grading scale: Pass/Fail

Literature: Multivariate image analysis. Geladi & Grahn ISBN 0-471-93001-6 and relevant articles in the field

Course period: 201702-201705

Send application to: Olle Hagman Olle.Hagman@ltu.se

Deadline for application: 170120

Contact person: Olle Hagman