It is very difficult for blind persons to do their daily work by their own. This paper propose a system to help blind persons read text labels and product packaging of hand held objects in their daily life. This system consists of camera for grabbing the image, processor for processing the grabbed image and headset for audio output. Blind user will only have to show objects in front of the camera and capture an image. The camera will grab the image. Input image given to the system is then converted into gray scale and apply SIFT (scale invariant feature transform) features and KNN (k nearest neighbor) algorithm on the image to detect text. Algorithm recognizes text from the image and according to that voice output of the recognized is given to the blind user. If product label is not detected, system gives instruction that product is not found.
Prof. Dr. Bilal BİLGİN