Non Homogenous Assistive Text Reading From Objects For Blind Persons

Text information provides important clues for many image based application such as assistive navigation. However locating and retrieving non homogeneous text is challenging task. The goal is to provide a framework for non homogeneous assistive text reading from objects for blind person based on text orientation. Non horizontal text is recognized using key-points generated by Scale Invariant Feature Transform and individual text characters are recognized based on line detection on text characters along with Optical Character Recognition. Development of the framework is performed using the Image Processing Toolbox of MATLAB. The extracted text characters are given in speech through audio for blind persons.