Braille, recently, has come to be written for the instructions of the elevators and the public place for blind people. However, only about 10% of people who are blind can read Braille. This paper describes a Braille reader system using a cellular phone embedded camera for people who are blind. The system extracts images of Braille dots from a photo taken by the phone camera according to Braille dot features using image-processing methods. The Braille dot features are the size and shape of areas extracted in the photo. Then, the system detects Braille cells by grouping the extracted images of dots according to the positional relation of the dots. After that, the system translates the detected Braille cells by searching a look-up table with the relationships between a Braille cell and a letter or a code, then reads aloud the translated text. In the evaluation experiment of the system, the successful translation rate was 67.5%.