A BCI-Based System for Facial-Action Recognition

• Main Authors: Zih-Yang Jiang, 江子揚
• Other Authors: Jzau-Sheng Lin
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/81716672112470072771

碩士 === 國立勤益科技大學 === 資訊工程系 === 103 === In this paper, a wireless-manner and EEG-based BCI device was developped to extract electroencephalogram(EEG) signals for recognizing the facial actions. In the proposed system, the Emotiv EPOC headset was used to extract the EEG signals. It consists of 14-channel bio potential sensors with goldplated electrodes offer optimal positioning for accurate spatial resolution. Based on the international 10-20 locations, AF3, AF4, F7, F8, FC5, FC6, T7, and T8 were used to capture EEG signals. In the Emotiv headset, the sampling rate is 128 Hz on the output as well as the internal sampling rate is 2048 Hz with 1.95- Least Significant Bit (LSB) voltage resolution. There are voluntary six actions i.e. Natural, Teeth Clenching, Furrow Brow, Raise Brow, Smile Left, and Smile Right to be recognized. The brain activity is recorded in real-time and discovered patterns to relate it to facial-action states with a cheap off-the-shelf EEG headset Emotiv EPOC device. First, EEG signals are sequentially and extracted from headset and transmitted to a personal computer with a wireless manner. Largenumber of feature vector of EEG can be reduced by a Wavelet transform. Then the reduced EEG signals can then be classified into six clusters by means of SVM algorithm with Gaussian kernel function. The experimental results showed a promising correct rate for the facial-action recognition through the proposed BCI system.