Emotion Classification in Response to Tactile Enhanced Multimedia using Frequency Domain Features of Brain Signals

Area of Science:

• Human-Computer Interaction
• Neuroscience
• Multimedia Systems

Background:

• Traditional multimedia lacks immersive sensory feedback.
• Tactile feedback, such as temperature changes, can enhance user experience.
• Measuring emotional responses to multimedia is crucial for understanding engagement.

Purpose of the Study:

• To develop and evaluate tactile enhanced multimedia (mulsemedia) systems.
• To assess human emotional responses to mulsemedia using brain signals.
• To compare the effectiveness of different feature extraction methods for emotion classification.

Main Methods:

• Generated tactile enhanced multimedia by synchronizing hot/cold air effects with video clips.
• Recorded electroencephalography (EEG) data from 21 participants using a MUSE headband.
• Extracted frequency domain features from EEG data and classified four emotions (happy, relaxed, sad, angry) using a support vector machine.

Main Results:

• Emotion classification accuracy was 76.19% using frequency domain features.
• This accuracy is significantly higher than the 63.41% achieved with time domain features.
• Frequency domain features proved more effective for emotion classification in mulsemedia.

Conclusions:

• Frequency domain features are well-suited for emotion classification in tactile enhanced multimedia.
• EEG-based emotion recognition can provide valuable insights into user experience in mulsemedia.
• This research paves the way for more realistic and emotionally resonant multimedia experiences.