The main objective of the thesis work is to show how image and video quality studies can incorporate cognitive and affective elements when evaluating Quality of Experience (QoE) for audio and video content. The focus of the thesis is on perceived video quality when video is streamed via a wireless network, as this can give an irregular throughput. The latest definition of QoE, as stated in the Qualinet White Paper (2013), “is the degree of delight or annoyance of the user of an application or service. It results from the fulfillment of his or her expectations with respect to the utility and / or enjoyment of the application or service in the light of the user’s personality and current state”.
There are still quite some unknowns in this area, especially around what the exact link is between affective factors such as involvement, sensory perception, and the underlying technical parameters. This thesis therefore sets out a model which aims to clarify the relationship between perceived video quality and involvement. It also aims to supply an operational definition of involvement, as well as a measure which can be used in the bigger framework of Quality of Experience. Assuming that Human Influencing Factors (IFs) are important to determine people’s preferences, a measure which reflects people’s opinion about scene content is necessary. A literature study indicated that related existing measures are not applicable since they assume objects (in marketing), interaction with the content (User Experience, games and immersion) or virtual reality (presence). To develop a fitting measure, a psychometric approach was adopted. First, it was necessary to
devise an operational definition of involvement. This was done through concept mapping, and the end result is a model for involvement with audio/video content with 6 attributes: captivation, expressions of involvement, informative interest, relatedness, negative affect and disinterest.
Based on the developed model, an involvement questionnaire (inQ) was developed and tested. A first draft version was quickly tested for understanding and wording through cognitive interviews. The second version of the inQ was then placed online, where about 100 participants scored 9 different video scenes with the involvement questionnaire. Results were analysed with the help of exploratory factor analyses, and a third version of the inQ was created. The third version of the InQ was validated through an experiment where 100
participants were shown 15 one-minute scenes, 5 different fragments, with three versions each: a reference version, a version where temporal artifacts (e.g. jerkiness) were induced, and a version where spatial artifacts (e.g. blockiness or blurriness) were induced. The results of this test showed that there was a 2-way relationship between the inQ and the perceived video quality, meaning that scores on inQ can partially predict scores on perceived video quality and vice versa. The results also showed that involvement can be defined by three attributes: positive / negative effect, relatedness and internal captivation. Video content can be judged differently across these factors, which could help differentiating people’s preferences for temporal or spatial artifacts in the future.
To conclude, the work presented in this thesis has shown that it is possible to create reliable and valid measures for Human IFs. This was established through both an explorative data analysis and a confirmative analysis. Furthermore, it has also shown that involvement with audio/video content is a salient Human IF and an important characteristic for QoE. Hence, involvement should be taken into account whenever participants are asked to judge video or audio quality, as it may change their quality judgement.