A Framework for the Encoding of Multilayered Documents

Electronic publishing of material digitized using imaging and OCR calls for a special delivery format capable of reconstructing original documents in a well-usable electronic form. We present a framework for the universal encoding of multilingual image-on-text documents, enabling retrieval systems to text-search and highlight hits on original page images. A generalized format for representation of image-on-text allows for integration of different OCR engines and target format encoders. This framework's current implementation encodes multilingual content into DjVu and PDF. Performance has been evaluated with focus on file size and shown that overhead of adding text layers is small compared to advantages and that output is comparable to other systems.