Mobile text reader for people with low vision

Abstract

People with low vision have visual acuity less than 6/18 and at least 3/60 in the better eye, with correction. The limited vision requires them to enhance their reading ability using magnifying glass or electronic screen magnifier. However, people with severe low vision have difficulty and suffer fatigue from using such assistive tool. This paper presents the development of a mobile text reader dedicated for people with low vision. The mobile text reader is developed as a mobile application that allows user to capture an image of texts and then translate the texts into audio format. One main contribution of this work compared to typical optical character recognition (OCR) engines or text-to-speech engines is the addition of image stitching feature. The image stitching feature can produce one single image from multiple poorly aligned images, and is integrated into the process of image acquisition. Either single or composite image is subsequently uploaded to a cloud-based OCR engine for robust character recognition. Eventually, a text-to-speech (TTS) synthesizer reproduces the word recognized in a natural-sounding speech. The whole series of computation is implemented as a mobile application to be run from a smartphone, allowing the visual impaired to access text information independently.