| 摘要: | 言語為表達情緒以及意思一個最直接的方式。一般正常人可以輕易地藉由它來傳達訊息，但是對於意識清醒卻無法開口講話的漸凍人來說，是一件難以想像的夢想。如果能由電腦建立一套系統能夠幫助他們來跟外界做溝通，對於照顧他的家人以及醫師將是一個極大的幫 助。 本研究的目的在於建構一套“眼動英文字母表達系統”，這套系統是藉由眼球特定移動的軌跡。量測眼球轉動所造成的微弱電位差，包含水平訊號以及垂直訊號，藉由量測到的水平訊號以及垂直訊號將可以透過電腦軟體來做分析以及使用者所表達符號的辨識。 在本論文中，將使用眼電圖法(Electro-OculoGraphy, EOG)，透過電極貼片來擷取眼球移動所產生的訊號，由電腦做訊號的處理，並使用眼球震盪的方向以及霍普菲爾網路(Hopfield neural network , HNN)，來判斷使用者所表達的符號。在本系統裡共建立了26個字母、4個標點符號以及3個刪除符號，可讓使用者能表達字句。 Human language is a direct communication among human beings to express his/her emotions and thoughts. It is easy for most of us to express their thoughts through languages, but the ALS (Amyotrophic Lateral Sclerosis) victims, it will be a fantasy. If a computer system can help those patients to communicate with outside world, then it will be a great help for the families and doctors whose responsibility is taking care of them. My research is to build English Letter expression system based on Electro-oculography. When a user wants to express him/herself, he/she simply needs to revolve his/her eyeballs clockwise three times, and the system will be initiated. After finishing each alphabet, if the users take a glance at a point one second, the system will recognize that the users have finished one alphabet; if the users are done with “writing,” he/she just needs to gyrate his/her eyeballs counterclockwise three times, and the system will be at a standby mode; the users can activate the system at any time. In my research, I used electrode to get both horizontal and vertical signal from eyeballs. Once the signals were picked, it will be magnified 334 times by Differential Amplifier, and then signal will be transmitted to the computer by analog-digital converter. After that, the computer will manage the signal and the recognize alphabets. The signal management includes: filtering disorderly signals, classifying codes, and detailed sorting for Hopfied Neural Network. In this system, I established 33 symbols to facilitate users, including 26 Latin alphabets, four commonly used punctuation symbols (space, comma, question, and dot), and three symbols (cancel letter, cancel word, and cancel sentence). Users will be able to communicate with and express to others easily using those 33 symbols. |
