Sistemas de Visão Industrial

C and MATLAB, Mathematical Analysis, Linear Algebra

Lecture method: an explanatory method in which facts, concepts, principles and generalisations are defined and presented by the lecturer and discussed with the class, followed by illustrative examples;

Experimental method: an active method in which students develop their knowledge through problem-solving and project work, involving group dynamics as well as individual and reflective work.

Completion of a series of practical assignments using Matlab to demonstrate concepts and provide examples of the application of specific techniques.

Carrying out a series of laboratory projects using computer vision hardware in groups of two students and employing a variety of tools: Matlab, C, Python, OpenCV and Deep Learning.

To recognise the importance of computer vision in current and future engineering challenges, in the fields of industry and robotics, security, agriculture, the environment, medicine, sport, mobility and intelligent transport systems, the IoT, amongst others;
To understand the principles of image formation, acquisition and representation;
To understand and apply image processing techniques in the spatial domain;
To understand the basic techniques for the representation and processing of features, colour and texture;
To develop industrial vision applications;

Learning outcomes and generic skills:
The ability to use image acquisition and processing devices to develop industrial vision solutions;
The ability to identify, propose and implement vision solutions in industrial applications.
To design, use and apply computer vision systems in industry for property evaluation, monitoring and control;
Ability to design and implement solutions in industrial contexts and in applied scientific fields.
Introduction to and motivation for research into vision applications within the scope of Industry 4.0 and using Deep Learning.

The rationale for computer vision in various fields of engineering and in industrial applications.
Objectives of computer vision in industrial applications and robotics;
Image formation and acquisition;
The Human Visual System: image formation, trichromatic theory and colour spaces.
Sensors, digital image formation and representation;
Analysis of binary, greyscale and colour images;
Low-level image processing in the spatial domain;
Filtering in the spatial domain;
Differential operators;
Morphological operations;
Colour and texture representations;
Region and contour descriptors.
Image recognition: models and patterns.
Lighting issues and techniques.
Industrial machine vision systems.
Other laser-based perception systems: 1D and 2D systems;
3D sensors and other forms of perception.
Image processing using deep learning.

Regular assessment and examination.

Component 1: Successful completion of assignments and projects: 25% of the mark.
Component 2: Two tests throughout the semester: 75% of the mark.

The first test will cover the concepts, techniques, development and implementation of the assignments and projects carried out from the start of the course until one week before the test.

The second test will cover the concepts, techniques, development and implementation of the assignments and projects carried out from the first test until one week before the test.

Sitting each test requires the submission of the corresponding assignments and projects.
Assessment by means of tests requires attendance at at least 75 per cent of the classes.

Date of the first test: 7th week of classes.
Date of the second test: the final week of the term.

Conditions for admission to the exam:
Attendance of at least 75 per cent of all classes.
Having successfully completed the practical assignments and projects.

The exam will cover the topics from the two periodic assessment tests.

NAO