情緒模型之建立方法及以該情緒模型進行的情緒偵測方式

專利類型

發明

專利國別 (專利申請國家)

中華民國

專利申請案號

102107320

專利證號

I 501093

專利獲證名稱

情緒模型之建立方法及以該情緒模型進行的情緒偵測方式

專利所屬機關 (申請機關)

國立雲林科技大學

獲證日期

2015/09/21

技術說明

本發明提出了一套系統，透過心電圖(Electrocardiogram, ECG)、皮膚電位反應(Galvanic Skin Responses, GSR)、血流量(Blood Volume Pulse, BVP)及脈搏(Pulse)四種生理訊號去分類悲傷(Sadness)、害怕(Fear)及生氣(Anger)三種情緒，當受測者的情緒狀態處於負面時，提醒受測者注意目前的情緒狀態，並做適當的調適。首先，從原始訊號計算出時間域、頻率域及非線性域的特徵，再透過循序前進浮動搜尋(Sequential Floating Forward Selection, SFFS)及F-score特徵選取演算法，選出較重要地特徵。最後，利用這些較重要的特徵，透過支援向量機(Support Vector Machine, SVM)分類器去訓練出負面情緒模型，即可馬上辨識出受測者的情緒。 Emotion is a great source of information in communication and interaction among people. Physiological responses are non-autonomic nerves in physiology. Physiological responses and the corresponding signals are difficult to control when a person is overcome with emotion. Therefore, this invention analyzes emotional physiological signals and constructs various negative emotional models. Accordingly, subject’s negative emotions can be detected by the negative emotional models. This invention collects and analyzes four physiological signals from subjects, namely electrocardiography, galvanic skin response (GSR), blood volume pulse, and pulse. Support vector regression (SVR) was used to train three trend curves of three emotions (sadness, fear, and anger). This invention has a high recognition rate of 90.6%.

備註

連絡單位 (專責單位/部門名稱)

智財管理組

連絡電話

(05)5342601轉2521

網址