표제지
목차
Abstract 11
1. 서론 13
2. 연구 방법 16
2.1. 연구 설계 16
2.1.1. 연구 개요 16
2.1.2. 가설 19
2.2. 데이터 수집 20
2.2.1. 시각 자극의 제시 방법 20
2.2.2. 아로마 향 22
2.2.3. 시각적 효과 25
2.2.4. EEG 측정 28
2.2.5. 실험 방법 32
2.2.6. 설문조사 36
2.3. 감정 분류 39
3. 결과 분석 45
3.1. 주관적 평가 45
3.1.1. Pleasant Scents 45
3.1.2. Unpleasant Scents 52
3.2. EEG 채널 수에 따른 평가 57
3.2.1. 채널 1개 사용 57
3.2.2. 채널 2개 사용 58
3.2.3. 채널 3개 사용 60
3.2.4. 채널 4개 사용 63
3.3. DWT 특징 추출에 따른 평가 66
4. 결론 68
4.1. 결론 68
4.2. 연구의 한계 및 향후 연구 방향 71
참고문헌 72
Table 1. Presentation methods by AR particles color and emission direction 20
Table 2. Pleasant scents images and particle colors 24
Table 3. Unpleasant scents images and particle colors 24
Table 4. Participant's information 40
Table 5. The RM ANOVA results of valence for pleasant scents 46
Table 6. Comparison by the response with significant differences of valence for lavender and rose scent 48
Table 7. The RM ANOVA results of arousal for pleasant scents 50
Table 8. Comparison results by the response with significant differences of arousal for the rose scent 51
Table 9. The RM ANOVA results of valence for unpleasant scents 53
Table 10. Comparison results by the response with significant differences of valence for the tar scent 54
Table 11. The RM ANOVA results of arousal for unpleasant scents 55
Table 12. Comparison results by the response with significant differences of arousal for tar scent 56
Table 13. Accuracy results of classifying using Fp1 and Fp2 channels 59
Table 14. Accuracy results of classifying using F3 and F4 channels 59
Table 15. Accuracy results of classifying using Fp1, Fp2, and F3 channels 61
Table 16. Accuracy results of classifying using Fp1, Fp2, and F4 channels 61
Table 17. Accuracy results of classifying using Fp1, F3, and F4 channels 62
Table 18. Accuracy results of classifying using Fp2, F3, and F4 channels 62
Table 19. Accuracy results of classifying using 4 channels 63
Table 20. Accuracy results of classifying by extracting features 66
Table 21. The number of data per class 67
Fig 1. Humans have five Senses 14
Fig 2. Research pipeline 18
Fig 3. The four visual stimulus presentation methods (methods 2 to 5) 21
Fig 4. Scent-related equipment used in experiments 22
Fig 5. How to locate scent emissions using a QR code 25
Fig 6. Application scenes 26
Fig 7. The brain is subdivided into 5 lobes 28
Fig 8. International 10-20 system for EEG 29
Fig 9. Locations of 4 EEG electrodes 30
Fig 10. Bio-signal related equipment used in experiments 31
Fig 11. Sequence of presentation methods by group using a Latin square design 32
Fig 12. Sequence of scents by participants using the Latin square design 33
Fig 13. Experiments with scents and AR particles 34
Fig 14. Russell's arousal-valence model of emotion 36
Fig 15. The Self-Assessment Manikin(SAM) used in the experiment to rate the evaluation of valence and arousal. 37
Fig 16. Binary classification based on a scale of 1 to 9 (low:1-5, high: 6-9) 41
Fig 17. 10-fold cross validation 41
Fig 18. Raw EEG signals of rose and tar scents 42
Fig 19. Band-pass filtered EEG signals of rose and tar scents 43
Fig 20. The average value of valence for pleasant scents 45
Fig 21. The average value of arousal for pleasant scents 49
Fig 22. The average value of valence for unpleasant scents 52
Fig 23. The average value of arousal for unpleasant scents 54