【摘要】 目的:通过对志愿者观看3D影片之后的脑电信号进行主成分分析,选取最能代表立体视觉疲劳度的主成分,运用BP神经网络对疲劳等级进行建模,提高对疲劳度等级的预测准确度。方法:采集15名志愿者观看五部不同3D影片前后的脑电信号,先对脑电信号进行疲劳度分级并选取特征通道;再对特征通道的脑电信号进行主成分分析选取影响最大的特征主成分,利用BP神经网络进行建模,根据建立的模型对立体视觉引起的疲劳等级进行预测,将预测结果与已知的疲劳等级进行对比。结果:根据文献中的疲劳等级将实验结果分成三个等级;据累计贡献率超过90%选取的前四个主成分建立的预测模型,准确度达95.4%。结论:运用主成分分析和BP神经网络的方法对立体视觉疲劳度进行预测,预测准确度较高,与直接根据脑电特征参数建立模型的方式相比简便和准确,这一方法对立体视觉引起的疲劳度分级及预测提供了新的思路。更多还原

【Abstract】 Objective: Principal component analysis(PCA)was performed on the EEG signals after the volunteers watched the 3D film. The principal components that can best represent the stereoscopic visual fatigue were selected, and the BP neural network was used to model the fatigue level to achieve the analysis of the EEG signal. Methods:Fifteen volunteers’ EEG signals were acquired before and after watching five different 3D movies,then grating the EEG signals fatigue using existing models as an objective basis. This paper used the method of PCA to analyze EEG signals of the characteristic channels that can greatest influence on fatigue.The BP neural network was used to model the characteristic principal component parameters, and the fatigue level caused by stereo vision was predicted according to the established model.Results:According to the existing fatigue level formula, our experimental results are divided into three fatigue levels.After the PCA, the first four principal components whose cumulative contribution rate exceeds 90% were selected;then using BP neural network to build a predictive model,and the prediction accuracy of the model is 95.4%.Conclusion: The PCA and BP neural network were used to establish prediction model of the stereoscopic fatigue, and we got a high prediction accuracy. It is simpler and more accurate than the previous method of analyzing EEG characteristic parameters and rebuilding models.And this result provided a new idea for the prediction of fatigue level caused by stereo vision.更多还原