To improve the navigation and orientation possibilities of blind or partially sighted persons, an assistant system named as Örientation Aid for the Blind Peop/e" is to be developed in the Institute for Visualization and lnteractive Systems (VIS}. The system is more than a navigation aid for the blind people, also provides further functions, which can help the blind to be able to function as normal persons. One of these functions is the text recognition. In order to realize the text recognition function, a module will be developed in this project and integrated into the existing system. This module firstly abstracts the text regions from the source image by using a Gabor filters based method and K-Means clustering method, and then the Otsu's method and A double threshold method will be used to binarize the abstracted text regions. Finally the Tesseract OCR engine recognizes the text from the binarized text regions. The theories and implementations of these in this module used technologies will be in this thesis described in detail. How to combine this module with the existing system and the results of this project will also be presented.