基于HOG特征的脑电地形图分类算法研究

BEAM Classification Algorithm Based on Histogram of Oriented Gradient Features

脑电地形图(Brain Electrical Activity Mapping,BEAM),是一种先进的研究脑功能和临床诊断的重要手段,既能进行病理诊断又可进行功能诊断,具有较高的敏感性。通过BEAM判断人在不同高低负荷下的疲劳情况并进行有效分类,能最大程度避免高危从业人员的危险发生。目前,大多数脑力负荷分类方法只是简单地利用脑电信号的四种频段特征进行分类,但分类效果并不理想。在此基础上,提出将脑电信号可视化分析,构建脑电地形图,并将方向梯度直方图(Histogram of Oriented Gradient,HOG)特征应用到BEAM分类中。BEAM是根据各频段功率谱密度值用不同颜色表示的球面头皮展成的平面图形,所以针对BEAM的分类研究是基于图像浅层特征的,而HOG正是图像处理中一种简单有效的浅层特征描述符。在BEAM中,HOG能有效地提取各脑功能区的边缘结构特征,并且能提取到地形图表象和形状的方向分布。首先将采集到的脑电信号进行预处理后,选择三频段脑电特征构建BEAM,进行HOG特征提取及分类任务,并与其他算法进行对比。得到的脑电地形图分类结果表明,提取HOG特征的BEAM分类精度最高,明显好于其他研究算法。

At present,most of the mental workload classification methods simply use the four frequency band features of EEG signals to classify,but the classification effect is not ideal.On this basis,this paper proposes to visually analyze EEG signals,construct BEAM,and apply the Histogram of Oriented Gradient(HOG)to the BEAM classification.BEAM is a planar graph formed by spherical scalps represented by different colors according to the power spectral density values of each frequency band.Therefore,the classification research of BEAM is based on the shallow features of images,and the HOG is a simple and effective shallow feature descriptor in image processing.In the BEAM,HOG can effectively extract the edge structure features of each brain functional area,and can extract the directional distribution of the image and shape of the topographic map.