Word Retrieval System for Ancient Arabic Manuscripts

摘要

Many museums and libraries that have archives containing historical Arabic manuscripts are beginning to digitize their collections to enable researchers and the general public access to them. Most of these collections are available in digital format with no annotation, limiting researchers to fully utilize them efficiently. This paper reports on a system we developed for word retrieval in Historical Arabic Handwritten manuscripts. Unlike other systems, this system does not use Optical Character Recognition or word spotting techniques. The system performs three different steps: segmentatlon, annotation, and word search. We designed a new interface to read and query Historical Arabic Handwritten manuscripts using text queries. The queried word is then highlighted in the page image of the manuscript.