A nonlinear quality of experience model for high dynamic spatio-temporal mulsemedia

Jalal, Lana;Murroni, Maurizio

2017-01-01

---

Abstract

In multimedia, quality of experience (QoE) accounts for the degree of delight or annoyance of the user of an application or service. Although humans have five senses, only two of these senses (i.e., sight and hearing) are stimulated by traditional multimedia contents. Therefore, the research efforts try to provide realistic media contents to the users. Realistic media contents are media with multiple sensorial effects, called mulsemedia, aimed at increasing user's experience through the five senses representation. This introduces a number of new issues, such as the QoE assessment for audiovisual sequences enriched with additional sensory effects such as light, wind, vibration, scent. QoE evaluation is based on mean opinion score (MOS) subjective tests measurement campaigns, which are time consuming, although allowing for the definition of statistical prediction models. The state of the art of QoE model for mulsemedia is based on parametric models. A linear model has been proposed, based on least square (LS) parameter estimation and validated on high dynamic spatio-temporal mulsemedia. This paper proposes a nonlinear model for predicting the QoE for high dynamic spatio-temporal mulsemedia. The parameter estimation relies on metaheuristic particle swarm optimization (PSO) which has been efficiently applied to optimization of nonlinear problems. A comparative analysis of the performance of the proposed model with the state of the art linear model for the QoE has been carried out based on the same MOS data set and assesses the effectiveness of the former with respect to the latter in case of high dynamic spatio-temporal mulsemedia.