The recent “resurrection” of interest in Virtual Reality has stimulated interest in the quest for true “immersion” in computer-generated worlds. True immersion may only ever be achieved through advanced BCI systems, but, until that day arrives, it is important to understand how it may be possible to measure human engagement and emotions within virtual worlds using psychophysiological techniques. This study aims to design an affective computing system, capable of responding to human emotions, within virtual environments. Based on the development of a Valence/Arousal model, a controllable affective VR, capable of evoking multiple emotions, has been constructed. Multiple variations of the VR have been evaluated subjectively using over 68 participants. More objective, physiologically-based experiments have been executed, in which the EEG, GSR and heart rates of 45 participants have been recorded during exposure to the most powerful affective environments, identified in the earlier study. Multiple affective recognition systems have been trained and crossvalidated against 30 participants and evaluated using the other 15 individuals. The results suggested that the trained classifiers perform highly accurately in the training database, but achieve random classification accuracies in the new dataset. It was highlighted that the extreme performance attenuation is due to the high individual differences in participants’ physiological responses, in emotional experiences.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:731904 |
| Date | January 2017 |
| Creators | Moghimi, Mohammadhossein |
| Publisher | University of Birmingham |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://etheses.bham.ac.uk//id/eprint/7924/ |
Page generated in 0.0019 seconds