A Semi-automatic Support to Adapt E-Documents in an Accessible and Usable Format for Vision Impaired Users

Abstract

Abstract. Electronic material (e-documents, e-books, on line resources, etc.) represents an essential tool for continuous learning for print-impaired people, provided it is well-structured. To obtain accessible and usable e-content, specific requirements should be applied from the early beginning or used when adapting existing electronic formats. In this paper we present a method, and a first associated prototype, for making e-documents in a format, which is accessible and usable for vision impaired users. The resulting environment is composed of various transformations, with different degree of automation, and applies a number of guidelines that have been defined for this purpose.

Semi-automatic generation of Arabic digital talking books

“…For a person who has visual impairment or a reading disability it is not possible to read printed documents [1]. The solution sometimes is to present these documents in Braille encoding which is expensive to produce, and not all persons with visual impairment can read Braille.…”

A Book Reader Design for Persons with Visual Impairment and Blindness

“…Studies in [25,26] are rather unique as they propose to convert e-books or e-documents available in PDF into formats that are easily accessible by persons with visual impairments (i.e., Braille, audiotape, or large print format). Understandably, descriptions of figures and graphs must be added manually in a semiautomatic process [25].…”

“…Studies in [25,26] are rather unique as they propose to convert e-books or e-documents available in PDF into formats that are easily accessible by persons with visual impairments (i.e., Braille, audiotape, or large print format). Understandably, descriptions of figures and graphs must be added manually in a semiautomatic process [25]. In a sophisticated image analysis strategy and seeking a higher reading accuracy, the study reported in [27] integrates the so-called iconic model (character-image classifier) and a linguistic model (word occurrence probability) with mutual entropy being the measure that assesses the disagreement between the two models (a high mutual entropy would mean a disagreement…”