融合脑血氧饱和度及面部表情的多诱因非合适驾驶状态识别

发布时间：2018-07-04 12:24

本文选题：多诱因非合适驾驶状态 + 前额叶脑血氧饱和度　；参考：《中南大学》2013年博士论文

【摘要】：摘要：随着人类社会的老龄化及文化发展的多元化,交通事故的主观诱因日趋复杂,这些诱因不仅包含传统意义下的疲劳及分神行为,同时包括饮酒、吸毒、疾病、药物等新型诱因。这些诱因致使驾驶主体进入多诱因非合适驾驶状态(Various Inducement Unfitted Driving States,VIUDS),降低了对驾驶环境的认知能力,并最终导致交通事故。目前,传统的方法只能实现单一诱因的检测,而无法在统一的技术构架下实现多种诱因的同时检测。针对这一问题,本文以生物医学及认知科学的最新研究成果为基础,在分析了多种主观诱因对驾驶主体中枢神经系统认知能力负面共性影响的基础上,选取了面部表情及脑前额叶血氧饱和度作为基本特征量,通过贝叶斯网络实现了认知能力的融合评估,并最终在同一技术构架下实现了VIUDS多诱因的统一检测。本文的主要创新如下： 1、在分析交通事故多种主观成因的基础上,得出了在多诱因作用下VIUDS本质上是中枢神经系统认知能力下降的状态；在分析基于单特征量及现有检测方法局限性的基础上,选取了面部表情及脑血氧饱和度作为反映认知能力的基础特征量,为认知能力的融合评估奠定了基础； 2、提出了一种认知能力的评估方法,该方法首先采用去趋势波动分析的方法计算脑额叶血氧饱和度时间序列的标度指数(Cerebral Oxygenation Saturation Scaling Exponents,COSSE),并通过该指数实现了主体中枢神经系统认知能力的分阶评估； 3、提出了一种基于石英音叉及椭球面声聚焦光声传感结构的脑额叶血氧饱和度检测方法。在检测过程中,额叶组织及石英音叉分别位于椭球面的两个焦点,在光激励脉冲的刺激下,焦点处额叶组织产生的弱光声信号经椭球面聚焦后被另一焦点处的石英音叉以共振的方式拾取,最后根据激励光强、光声信号强度及血氧饱和度之间的定量关系,可实现前额叶脑血氧饱和度的远场无创测量； 4、提出了一种正交流形保持投影的表情方法。针对流形保局投影方法的局限性,即不能在低维投影空间保持流形全局结构的不变性及投影矩阵非正交的问题,重新定义了目标约束函数,并采用格拉姆-施密特正交化过程重新求解投影矩阵,不仅在投影空间有效地保持了流形的全局及局部结构,同时有效地降低了特征维数,提高了表情识别的准确率； 5、提出了一种基于贝叶斯网络的认知能力多特征融合评估方法,该方法采用脑额叶血氧饱和度时间序列的标度指数、疲劳表情、眨眼的频率等作为融合的基本特征,通过贝叶斯网络计算了这些独立随机量对应的后验概率,并采用后验概率作为认知能力的评估依据,从而有效地克服了采用单特征进行检测的局限性,提高了评估的精度。论文在最后一章对全文工作进行了总结,并指出了后继的工作及研究方向。本文包括图33幅,表10个,参考文献159篇。
[Abstract]:Abstract: with the aging of human society and the diversification of culture, the subjective causes of traffic accidents are becoming more and more complex. These inducements not only include fatigue and distraction in the traditional sense, but also include drinking, drug taking and disease. New inducements such as drugs. These inducements cause the driving subject to enter the multi-inductive unsuitable driving state (VIUDS), which reduces the cognitive ability of driving environment and eventually leads to traffic accident. At present, the traditional method can only realize the detection of single inducement, but can not realize the simultaneous detection of multiple inducements under the unified technical framework. Aiming at this problem, based on the latest research results of biomedicine and cognitive science, this paper analyzes the negative common effects of various subjective inducements on the cognitive ability of central nervous system of driving subjects. The facial expression and prefrontal blood oxygen saturation were selected as the basic characteristics, and the fusion evaluation of cognitive ability was realized by Bayesian network, and finally the unified detection of multiple inducements of VIUDS was realized under the same technical framework. The main innovations of this paper are as follows: 1. On the basis of analyzing the subjective causes of traffic accidents, it is concluded that under the action of multiple inducements, VIUDS is essentially a state in which the cognitive ability of the central nervous system decreases; Based on the analysis of single feature quantity and the limitation of existing detection methods, facial expression and cerebral blood oxygen saturation are selected as the basic feature quantities to reflect cognitive ability, which lays a foundation for the fusion evaluation of cognitive ability. 2. A method for evaluating cognitive ability is proposed. In this method, the scalar index (Cerebral Oxygenation Satation scaling scale COSSE) of the frontal lobe blood oxygen saturation time series was calculated by the method of detrend fluctuation analysis, and the cognitive ability of the central nervous system was evaluated by the index. 3. A method based on quartz tuning fork and ellipsoidal acoustic focusing photoacoustic sensor structure is proposed to detect the cerebral frontal lobe blood oxygen saturation. During the detection process, the frontal lobe tissue and the quartz tuning fork were located in the two focal points of the ellipsoid, respectively, stimulated by the light excitation pulse. The weak photoacoustic signal produced by the frontal lobe tissue at the focal point is picked up by the quartz tuning fork at the other focal point by the ellipsoidal focusing surface. Finally, according to the quantitative relationship among the excitation light intensity, the intensity of the photoacoustic signal and the saturation of blood oxygen, The far field non-invasive measurement of cerebral blood oxygen saturation in prefrontal lobe can be realized. (4) A positive AC shape preserving projection expression method is proposed. In view of the limitation of the local projection method, that is, the invariance of the global structure of the manifold and the nonorthogonality of the projection matrix in the low-dimensional projection space, the objective constraint function is redefined. Using Gram Schmidt orthogonalization process to solve the projection matrix, not only the global and local structure of the manifold is effectively maintained in the projection space, but also the feature dimension is reduced effectively, and the accuracy of facial expression recognition is improved. 5. A Bayesian network based multi-feature fusion evaluation method for cognitive ability is proposed. The basic features of the fusion are the scalar index of the time series of cerebral frontal oxygen saturation, the expression of fatigue and the frequency of blink. The posteriori probability of these independent random variables is calculated by Bayesian network, and the posteriori probability is used as the basis for the evaluation of cognitive ability. Thus, the limitation of single feature detection is effectively overcome and the accuracy of evaluation is improved. In the last chapter, the thesis summarizes the work of this paper, and points out the following work and research direction. This paper includes 33 figures, 10 tables and 159 references.
【学位授予单位】：中南大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：U491.254;U492.8

【参考文献】