Haute définition

La haute définition (HD) désigne une classification d'équipements de télédiffusion et de vidéo numérique ayant une définition d'au moins 1280 par (720p), puis 1920 par 1080 pixels (1080i et 1080p) et par la suite, 2560 par 1440 pixels (1440p). Elle est l'évolution du SDTV. Le terme HD s'étend à l'ensemble des techniques audiovisuelles numériques telles que le HDV (grand public), le HDCam, la télévision (DVB-T, DVB-S, DVB-C), les supports disque Blu-ray, HD DVD, l'enregistrement multimédia sur disque dur, ainsi qu'au stockage de données informatiques. Principes techniques Les spécificités de la HD nécessitent une définition de l'image vidéo supérieure (d'un facteur de deux au minimum) à celle de la référence dite « SD » (pour Standard Definition). La SD présente une définition (en mode ) de en base par en hauteur, notée en abrégé . Les industriels considèrent qu'un équipement vidéo peut être qualifié HD dès lors qu'il traite ou exploite un signal vidéo dont la définitio

A Bayesian Method for Matching Two Similar Graphs without Seeds

Matthias Grossglauser, Pedram Pedarsani

Approximate graph matching (AGM) refers to the problem of mapping the vertices of two structurally similar graphs, which has applications in social networks, computer vision, chemistry, and biology. Given its computational cost, AGM has mostly been limited to either small graphs (e.g., tens or hundreds of nodes), or to large graphs in combination with side information beyond the graph structure (e.g., a seed set of pre-mapped node pairs). In this paper, we cast AGM in a Bayesian framework based on a clean definition of the probability of correctly mapping two nodes, which leads to a polynomial time algorithm that does not require side information. Node features such as degree and distances to other nodes are used as fingerprints. The algorithm proceeds in rounds, such that the most likely pairs are mapped first; these pairs subsequently generate additional features in the fingerprints of other nodes. We evaluate our method over real social networks and show that it achieves a very low matching error provided the two graphs are sufficiently similar. We also evaluate our method on random graph models to characterize its behavior under various levels of node clustering.

2013

Context-based Quality of Experience in Immersive Multimedia

Anne-Flore Nicole Marie Perrin

Since 1895, when the Lumière brothers came out with the projected cinematograph, motion pictures and more generally multimedia contents have considerably improved and are still experiencing a very dynamic development. Advances involved improving various modalities that influence the Human Visual System (HVS) or the human perception. Indeed, migrations concerning audio, resolution, color, brightness, depth representations and frame rate have been performed or are being implemented. A high-quality multimedia content targets at least an acceptable level of consumer¿s satisfaction. It assumes to be a combination of sensory stimuli whose goal is to entertain, to educate or to inform, and can convey an artistic intent. Measuring the end-user satisfaction is a challenge as it is a complex multi-dimensional concept. To understand, measure and predict user satisfaction, researchers focused on the perceived quality of multimedia contents. Quality is the expression of an individual comparison and judgment process. It is based on inherent characteristics of a stimulus. It turns out that network-related impairments, major distortions in content delivery, are not considered when evaluating contents¿ inherent features. The quality measure considering the entire service pipeline (including delivery) is referred to as QoS. A more user-centric quality concept has been created under the consideration that not only technical quality and delivery artifacts influence an experience. Accordingly, QoE represents the user¿s degree of delight or annoyance for an application or service in regards to his prior knowledge, expectations and current state. This work addresses the design of new methodologies for QoE subjective evaluation. The research includes identification, definition and investigation on influence factors of QoE. Possibility to predict QoE using physiological signals gathered during subjective assessment is also considered. This thesis is positioned as a framework for the assessment of emerging technologies. Its main focus is on HDR and Wide Color Gamut (WCG) representations, for these emerging technologies provide faithful and realistic content; immersive technologies, such as 360 imaging and VR gaming; as well as the combination of typical multimedia representations (e.g. Standard Resolution (SD), High Definition (HD) and 4K contents along with no audio, stereo and surround audio). From conducted experiments on HDR and WCG, important recommendations came out for these technologies¿ deployment regarding representation and compression strategies while indicating future directions for QoE assessment. We designed a new subjective methodology for 360¿ HDR contents. It investigates which dimensions should be included in immersive and realistic contents evaluation while considering representation constraints (e.g., no spatial pair comparison possible for 360¿ contents). Expectations and novelty effect evaluations, two emerging-technologies-related aspects that can bias QoE-centric subjective assessment, have been implemented in the context of Virtual Reality (VR) gaming. Finally, attempts to predict the SoP, an important QoE-related notion when analyzing physiological signals (brain activity (Electroencephalography (EEG)), heart activity (Electrocardiography (ECG)) and respiration) have been performed for various multimedia consumption scenarios (in terms of quality, resolution and sound systems as well as typical use cases).

EPFL2019

Context-based Quality of Experience in Immersive Multimedia

Anne-Flore Nicole Marie Perrin

EPFL2019

A Bayesian Method for Matching Two Similar Graphs without Seeds

Pre-interventional pulmonary hypertension predicts mortality in patients with aortic stenosis after transcatheter aortic valve replacement. A definition update

Context-based Quality of Experience in Immersive Multimedia

A Bayesian Method for Matching Two Similar Graphs without Seeds

Context-based Quality of Experience in Immersive Multimedia

Pre-interventional pulmonary hypertension predicts mortality in patients with aortic stenosis after transcatheter aortic valve replacement. A definition update