From Image to XML: Monitoring a Page Layout Analysis Approach for the Visually Impaired

摘要

Page layout analysis and the creation of an XML document from a document image are useful for many applications including the preservation of archived documents, robust electronic access to printed documents, and access to print materials by the visually impaired. In this paper, the authors describe a document image process pipeline comprised of techniques for the identification of article headings and the related body text, the aggregation of the body text with the headings, and the creation of an XML document. The pipeline was developed to support multiple document images captured by the head-mounted cameras of a reading device for the visually impaired. Both automatic and manual adaptations of the pipeline processed a sample of 25 newspaper document images. By comparing the automatic and manual processes, we show that overall our approach generates high-quality XML encoded documents for use in further processing, such as a text-to-speech for the visually impaired.