|OFICIAL||Electrical and Computer Engineering|
|Responsible unit:||Department of Electrical and Computer Engineering|
|Course/CS Responsible:||Doctoral Program in Electrical and Computer Engineering|
|Acronym||No. of Students||Study Plan||Curricular Years||Credits UCN||Credits ECTS||Contact hours||Total Time|
|PDEEC||6||Syllabus since 2007/08||1||-||7,5||70||202,5|
|Jaime dos Santos Cardoso|
This graduate course on Image Analysis and Recognition aims to give the student the ability to understand and apply some of the recent advances in this rapid evolving field of Image Analysis and recognition. There is a text book together with a list of selected original research papers in order to allow the students to follow the advances in the addressed topics. The course main topics will allow the students to gain the competences in: image segmentation, tracking, Image registration and object and pattern recognition and matching. The course will discuss the use of the learned methods and techniques in applications such as visual inspection, document processing, biomedical and biometrics.
The course main topics will allow the students to gain the competences in: image segmentation, tracking, Image registration and object and pattern recognition and matching. The course will discuss the use of the learned methods and techniques in applications such as visual inspection, document processing, biomedical and biometrics.
1. Image Enhancement a. Intensity, local and frequency methods b. Wavelets 2. From color to edges and textures a. Edge and corner detection b. Texture analysis 3. Segmentation a. Introduction b. Image Segmentation using basic clustering methods. Embedding local constrains. Mean Shift c. Segmentation by graph-theoretic clustering. Graphs. Affinity measures. Graph Cuts and normalized cuts. 4. Motion analysis a. Background subtraction b. Optical flow c. Tracking i. Tracking using linear dynamical models: Kalman filtering. (Examples: tracking people, tracking vehicles). ii. Tracking with non-linear dynamical models: Extended Kalman filtering and Particle filtering. 5. Image Registration a. Multi view geometry i. Projective transforms and invariance. ii. Photometric transforms and invariance. iii. Cameras and calibration. b. Strategies for image registration of rigid and non-rigid objects. c. Local invariant features and similarity measures.. 6. Image Recognition a. Machine learning tools b. Feature extraction and selection: i. Principal Component analysis. ii. Object and shape representation using invariant features. c. Object modeling i. Active appearance models. ii. Constellation model and the Implicit Shape Model. iii. Bag of visual term models. d. Recognition examples (face detection and recognition, pedestrian finding)
The course is organized in a set of 28 lectures together with oral presentation and computer vision labs.
|Trabalho de campo||30,00|
|Elaboração de projeto||40,00|
|Frequência das aulas||42,00|
|Trabalho de investigação||8,00|
Grading and evaluation is based on the following scheme: Assignments: 30% (10% for each one of the 3 assignments) Presentation of a selected research paper: (10%) Project: 30% Final exam: 30% Grading will be from 0 to 20. A PASSing grade corresponds to a minimum of 10.
The workload of the course consists of 3 assignments, presentation of one reading and a project. List of assignments: Assignment 1: Image Segmentation Assignment 2: Human Tracking in Videos Assignment 3: Image Registration/Recognition The assignments can be developed by groups of 2 students. However, the groups’ members must vary between assignments. A report with the full analysis of the applied methodologies and results must be presented by the students, which must include results, illustrations of results and a discussion of the results. Additionally, commented code (published html) must be presented of the methodologies’ implementation. The report must be presented in a pdf file. A small 10 minutes presentation of the performed work may also be requested. To each student will be assigned a reading of a selected research paper that will be presented during the classes during the semester. A project will be also assigned to a team of 2 students in the second half of the course. A report in IEEE PAMI format is required with a minimum of 6 pages and a maximum of 10 pages. Commented code and illustrative results must be reported in an additional pdf file. The projects will be presented in special sessions of the course. NOTE: it is recommended that the chosen project overlaps with the students’ existing specific interest areas. While the student may present a project of his/her choice the acceptance of such project requires the approval of the professors in charge. Late policy: The students must meet the deadlines. One week delay is allowed for the assignments and project with a 10% penalty on the corresponding grade.