Design of visual aid for blind people

Abstract

The proposed system is a design of visual aid for blind people. This design will help blind users to detect bank currencies and also aid them in reading printed words. It also helps them to identify human obstacles in front of them. For reading the printed texts, a combination of hierarchical optimization algorithm and pattern recognition in OCR is being used in our design. The characters from the text will be localized and isolated by pattern recognition. The resultant image of the character will be preprocessed using noise reduction filter. Text strings will be formed by grouping the identified characters from the characteristics extraction process. The output will be given as the speech for the corresponding text by converting the identified text to speech. Bank note identification is done by using the fast and efficient SIFT algorithm. It is precise and accurate. It will compare every input with the database templates and gives the highest match as output currency note for the blind user in their earphones. Obstacle identification is done using the Viola–Jones AdaBoost algorithm and using the same any obstacle can be detected and voice output is produced.