Alternative Product Label Reading and Speech Conversion: An Aid for Blind Person

An aid to help handicap people is proposed in this paper for reading text labels in their day to day life which can be executed by isolating the object through cluttered backgrounds or either surrounding object in the camera view. This paper establishes a productive and efficient motion based technique for defining a region of interest (ROI) in the video by shaking the object in the image. Gaussian based background subtraction method is used for extracting region of movement of an object. Text localization and recognition is done by using extracted ROI; since it is useful for retrieving text information. Learning gradient features based algorithm is used for text localization in an Adaboost model. Optical Character recognition technique is used for recognizing text characters in localized text regions. These text codes are then recognized and then converted into speech signal as output for visually impaired people.