A Real Time Text Detection&Recognition System to Assist Visually Impaired

Number of blind or visually impaired people in the world are constantly increasing due to reasons like diabetes, accidents, ageing etc. Along with navigation, they also needs assistance in reading various boards, labels etc. in their daily life. A reading assistance system based on text detection and recognition from natural scenes is proposed in this work. Camera based system detects text in natural images using Maximally Stable Extremal Regions algorithm. Detected text is recognized using Optical Character Recognition and converted into speech using Text-to-Speech synthesizer for blind user to listen through headphones. Proposed method is evaluated on ICDAR 2011 dataset.