Advanced techniques for digital imaging
Constantin Vertan
Coordinator
Overview
The Master's degree programme "Advanced Digital Imaging Techniques" (TAID) responds to the current requirements of development and evolution of the IT&C service economy, in the context of the generalization of digital image production and exploitation. The fields of activity covered are virtually unlimited, ranging from 'consumer' applications (digital camera technologies and mobile 'smartphone' terminals), medical (medical image analysis and processing products and technologies), military (satellite image processing products and technologies), security (surveillance and biometric systems), industrial automation (product inspection systems), robotics (human-machine interface systems) and many others.
The training is based on the application of programming, algorithmic and machine learning techniques.
Potential employers target both academic (teaching and research profile) and industrial R&D environments such as organisations/firms of any size, from small (e.g. start-ups and spin-offs) to multinationals.
Who is it for?
The Master's programme is mainly aimed at engineers in the fields of Computer and Information Technology, Electronic Engineering, Telecommunications and Information Technology, Applied Engineering Sciences, Systems Engineering. The programme can also be taken by graduates in mathematics, computer science or cybernetics.
Objectives of the Master's programme
The TAID Master's program aims to train graduate engineers to model and design image processing, image analysis and computer vision software/hardware systems for specific applications, as well as to identify and analyze specific problems and develop strategies to solve them.
Specialist skills offered to graduates
The competences recognised by the RNCIS and listed in the diploma supplement are:
- Thorough knowledge of design concepts, principles and methodologies specific to the fields of image analysis and processing and their applications;
- Ability to design and implement as well as test and evaluate complex image processing and analysis systems;
- Ability to create and implement mathematical models appropriate to specific image processing and analysis concepts;
- Ability to model and implement the software components of an image processing and analysis application for different systems (Windows/Android);
- Design and implementation of advanced database applications (data mining, database theory and design, including distributed, multimedia technologies).
Examples of research directions addressed
The titles of the dissertations in progress over the last two years are an enlightening example of the complexity and topicality of the topics covered in the TAID Master's programme:
- Automated algorithms for tracking people and their behaviour in video streams
- Digital image re-framing algorithms based on automatic content analysis
- Distributed application for image styling
- Detection of objects of interest in thermal images acquired with telephoto optical systems using deep neural networks
- Automatic analysis of facial expressions in children
- Duplicate detection of digital colour images in a multimodal database
- Facial analysis for automatic sibling recognition
- Recommendation system for a multimedia platform
- Distributed localization solution based on content similarity in images
- Automatic face rejuvenation/aging system
- Android app for displaying and transmitting information from digital mammography images
- Simulated expression detection
- Vehicle detection from natural images using convolutional networks
- Using the three-dimensional discrete cosine transform in video compression
- Method for tracking pedestrians in image sequences
- Detection of traffic lights (traffic lights, vehicles) in video sequences
- Detection of acquired digital images with "finger-in-front-of-lens" degradation
- Intelligent traffic light system for an intersection
- Facial analysis method for emotion recognition in image sequences
- Automatic traffic monitoring: traffic jam detection and accident prediction
- Comparative evaluation between SDKs for neural networks dedicated to mobile platforms
- Combining colors in clothing design and aesthetics using convolutional neural networks for colorization
- Efficient analysis of facial expressions
- Neural network optimization for integrated circuit applications
- Unsupervised information extraction method for improving emotion recognition in facial images
- Denoising and sharpening using U-Net architecture
- Study on the efficiency of a convolutional network to detect important points on the human face
- Facial expression transfer
- Facial recognition authentication service