隨著時代演進，知識日新月異進步，各種領域書籍資料大增，電子書因而被開發，但仍有大量書籍無電子檔存在。視障生閱讀輔助設備也逐漸成為研究的熱門領域之一，但這些研究都有一些缺點存在。例如：硬體設備價位太高、閱讀之文件必須為電子書；即使有電子檔存在，經由圖文分離再透過觸控螢幕可點選閱讀區域，但使用者仍不能隨心所欲選取書籍閱讀文字等缺點。因此我們提出一套新的閱讀輔助設備以提供視障者閱讀文件。本系統不需要電子檔的輔助，當閱讀時，只要用手指碰觸文件的文字即可。本系統分三部分執行，首先使用者將文件放置桌上的指定區域，利用kinect系統平台之彩色影像資訊擷取文件，進行圖文分離與文字辨識。接著，當盲生用手指碰觸文字欲閱讀時，利用Kinect系統平台之深度影像資訊取得手指位置；並與已建立之座標系匹配，以取得文字。最後透過文字轉語音或點字顯示器方式輸出。 As time goes on, a huge mass of progressive knowledge is developed and then the e-book is built for paper reduction and environmental protection. But electronic documents of many old dated of historical documents don’t exit. Research on blind reading aid device is a popular topic gradually, but there are some drawbacks on these methods. For example, hardware equipment is too expensive and reading format should be limited to e-document. It is still inconvenient for blind reading with e-document via the techniques of text graph separation and touch pad panel because he can’t read any specified region of the document as he wishes. A novel blind reading aid device is proposed without e-document and the user only need to point the document with his finger. This system is composed of three parts. First, we set up a specified region for document under the kinect sensor, and then text graph separation and character recognition is processed successively. Next, we use depth image of kinect sensor to get the coordinates of user’s finger when user wishes to read the document, and then match with constructed coordinate database to get the character. Finally, system output is obtained via text to speech or Braille display.
