Assistance vision for blind people using k-NN algorithm and raspberry Pi

Abstract

Eyes are the natural camera that human beings possess. But not all the people are lucky enough to have them. About 15 million people in India suffer with visual impairments. So, through this chapter, an attempt is made to help them by providing assistance for text reading from the documents and papers. The characters present in the text are isolated from cluttered background or the surrounding objects in the camera view. The characters are separated from one another by drawing contours. Once the text characters in the localized text regions are detected, they are binarized and trained using k-Nearest Neighbor (k-NN). The trained data is stored in a file and these will be the classifiers. These classifiers are loaded while testing. The text in the image is recognized based on the loaded classifiers file. This recognized text is stored and converted to speech.