基于眼电的智能输入系统研究

为了帮助肢体运动功能障碍患者与外界交流，设计一套基于眨眼眼电的便携式智能输入系统，使用者仅通过眨眼便能控制人机交互界面上的虚拟键盘进行字符输入。该系统中的信号采集模块先对眼电信号经行预处理，然后将其转换为数字信号传输至微处理。微处理首先通过数字形态学方法滤除信号中的尖峰噪声，然后通过动态阈值和归一化、微分算法识别主动眨眼信号，并完成对虚拟键盘的控制，实现中文、英文或数字的输入。选择12名受试者分别在常规实验室、电磁干扰较大的磁共振设备室和户外的运动场进行测试。结果表明，该系统在上述3个不同的环境中能够准确识别眨眼信号并实现中文、英文及数字的输入，而且输入字符的平均准确率不低于98%，中文、英文字符和数字的平均输入速度分别为(2.8±0.3)、(6.6±0.35)、（9.7±0.38）个/min。因此，该系统的抗干扰能力较强，能够帮助肢体运动功能障碍患者实现与外界交流。

Research of Intelligent Speller System Based on EOG

A portable intelligent speller system based on blink electro oculogram (EOG) was designed to help patients of limb movement function disorders to communicate with other people. The user can control virtual keyboard to spell characters by blinking on human-computer interaction (HCI) interface. The signal would be pre-processed by acquisition module of the system. Then the signal was converted to digital signal, and the digital signal was transmitted to microprocessor. Pike noise of the digital signal was firstly filtered by mathematical morphology. Then dynamical threshold algorithm and normalization, differential algorithm were used to detect the blink. According to the information of EOG, virtual keyboard was controlled to spell characters by microprocessor. In this paper, 12 subjects were tested the system in three occasions (routine laboratory, magnetic resonance imaging equipment room with the large electromagnetic interference, and outdoor playground). Results showed that the system could accurately identify blink signal and spell characters including Chinese, English and numbers. Besides, the mean average precision of input character was above 98%. The average input speeds of Chinese character, English character and number were 2.8±0.3, 6.6±0.35 and 9.7±0.38 (per/min), respectively. Hence the system which has higher interference rejection is appropriate for patient of limb movement function disorders.