中文摘要：

目前基于脑机接口的脑电信号研究得到越来越广泛的关注,然而传统脑电信号采集需要使用电极帽并涂抹胶泥膏而不被大多数用户所接受.因此,将研究使用无需涂抹胶泥膏的独立电极采集脑电信号,然而使用独立电极采集脑电信号容易出现干扰大、信号不稳定等缺陷.为了快速有效提取脑电信号特征并克服独立电极采集脑电信号的缺陷,将采用低通滤波方法进行工频干扰的滤除,利用独立成分分析（ICA）实现脑电信号中的眼电伪迹分离,并在此基础上通过设置水平眼电和垂直眼电阈值以及各个独立成分在脑部位置的空间分布特性实现眼电伪迹的识别.最后,分别利用β波能量以及样本熵来衡量人脑专注度的高低,仿真结果表明两者均与专注度成正相关,实验以NeuroSky专注度为基准,将两种算法分别与其进行对照.此外,样本熵与NeuroSky算法的相关度比β波能量法提高了26％,说明样本熵专注度提取算法更能精确跟踪人脑注意力的变化,对脑电信号专注度的衡量与实际更加吻合.

英文摘要：

The study of electroencephalogram signals based on brain-computer interface gets more and more attention, but the acquisition of the electroencephalogram signals by using traditional electrode cap cannot be accept- ed by most users because of the use of electrode cap and cement. So we acquire electroencephalogram signals by using independent electrode. However using independent electrode to acquire electroencephalogram signals makes big interference, and the signal is not stable. In order to quickly and efficiently extract brain signal characteristics, low pass filtering is used to remove power frequency interference, and we use independent component analysis （ICA） to realize the separation of eye artifact and electroencephalogram signals. Then we achieve the identification of eye artifact through the horizontal eye electrical threshold settings, vertical eye electrical threshold settings and spatial distribution characteristics of individual components in the location of the brain. Then we can use the energy of β-band and sample entropy to measure the concentration. Simulation results show that both the energy and entro- py are positively related with the concentration. Experimental measurement is on the basis of patented algorithm of NeuroSky. Compare the two algorithms with the algorithm of NeuroSky respectively, the results show that the corre- lation coefficient between the entropy and the value of attention of NeuroSky increases 26% comparing with the cor- relation coefficient between the energy and the value of attention of NeuroSky, which indicates that sample entropy can accurately track the change of attention and has a better effect in the feature extraction.