An Intelligent Spell System Based On EOG

Patients with motor neuron disease lost their functions of each motion systems,including language features, however, their can still control their eyes freely. This kind of patients have lost the ability of communicating with outside world, at present,auxiliary tools that can help them to communicate with outside world are scanty. If there is possibility of designing a set of human-computer interaction(HCI) system that controlled by eye movement, then, we can help those patients to achieve normal communication with outside world. This paper took this as a breakthrough point,studied and designed a set of auxiliary system of disabled based on Electro-oculogram(EOG).Users can control the HCI interface of the system to spell characters easily only by blinking, so they can achieve the goal of communicating with outside world. This system supports three kinds of input methods including Chinese, English and Arabic numerals, and it can display EOG waveform of users in real-time. The system’s hardware adopts detachable modular design. The data acquisition module uses analog front-end ADS1299 of Texas Instruments(TI). The system control module uses S3C6410 as the CPU, the HCI interface is displayed on a 10.1 inches LCD screen. The system uses two lithium batteries to provide power for the aim of portable. In the aspect of software, the system modified and transplanted Bootloader, cut and transplanted Linux kernel, made and transplanted yaffs2 file system. Then, the system developed SPI driver and Qt application. The system completed the construction of hardware to software.Compared with the similar devices on the market, this system has the following characteristics. Firstly, this system set up a intact HCI system. Secondly, the users could set their own threshold of blink EOG, which will make a more accurate spelling. Thirdly,this system has a simple operation, users can easily control the system only by blinking.Lastly, This system has a small volume, and it uses lithium battery for power supply,which will make it very convenient to carry around.In order to test the performance of the whole system, six subjects conducted spelling test in different circumstances. The test result shows that the system could accurately identify subjects’ blink action, then the blink action will be served ascommand to control the HCI interface to spell. The average accuracy rate reached99.3%. The test demonstrated that this system can be a excellent auxiliary tool to help those disables to communicate with the outside world.