Fondamenti di elaborazione del segnale multi-dimensionale (Multi-dimensional signal processing)

a.a. 2011/12 (academic year 2011/12)

• Information on the course
  • Assessment
  • Syllabus
• Course materials
  • Slides
• Useful links

Information on the course

Assessment

The course proficiency will be evaluated on the basis of a written project that each student have to submit to the course teachers.

The project will consist in a research in the literature on an applicative or theoretical problem related to the topics covered in the course. The student have to produce a short report in which the problem is analyzed and the solutions found in the literature are summarized and discussed. Optionally, the project can be extended to include an implementation of the solution.

The research problem to be developed will be assigned by the teachers, in agreement with the student, choosing one of the following topics:

• Image fusion through multiscale representation: When the depth of field of the acquisition setup cannot cover the depth of the scene, multiple images with different focus setup can be acquired and fused for obtaining a globally focused image of the whole scene.
• Image registration through multiscale representation: Images that partially describe the same scene, taken from slightly different points of view, can be stitched together for representing the whole scene. Solution of subproblems such as identifying matching reference points or the optimal transformation can be benefit of a multiscale representation.
• Independent Component Analysis: In the course, several applications employing Principal Components Analysis (PCA) were described. Independent Components Analysis is a further technique also used to decompose a signal into multiple subcomponents. Study its applications to image processing.
• Segmentation of individuals in surveillance indoor application: The segmentation of individuals in surveillance is the first and critical step for subsequent complex tasks such as tracking and recognition. The usage of multiple views of the same scene can be a valid approach to achieve a robust segmentation of individual in indoor conditions.

Once the topic has been chosen, the student has to inform the teachers by email (stefano.ferrari@unimi.it, fabio.scotti@unimi.it, roberto.sassi@unimi.it), indicating the title of the report, and possibily illustrating the subproblem or the application field that will be considered, or any other research criteria that will narrow the scope of the report. The teachers, after evaluating the relevance and the appropriateness of the proposed work, will assign the project title to the student.

Syllabus

Course materials

Slides

Useful links

• http://www.imageprocessingplace.com/
  textbook web site (Gonzalez, Woods, Digital image processing)
• http://hpe.pearson.it/gonzalez
  Italian edition textbook web site (Gonzalez, Woods, Elaborazione delle immagini digitali)

Home page     Didattica (Teaching)