Abstract

Ancient Egyptian hieroglyphic writing was used productively for over 4000 years. But even after knowledge of this writing system was finally lost, the pictographic script fired the imaginations of historians and travellers for hundreds of years. Those would-be translators imbued the script with all sorts of mystical meanings that bore little or no relation to their true contents.

In the nineteenth century, Egyptian hieroglyphs were finally deciphered; a discovery which led to sparked off passionate interest in the region that lasted well into the twentieth century. In this time, great inscriptions were copied down by hand and published for dissemination to a wider audience. Despite advances in typesetting, many of these texts continue to be reprints of the original handwritten documents, making them relatively inaccessible for computer-based analysis.

In order to change this, research is being conducted primarily into sketch recognition technology that could serve to transform these texts into computer-readable documents. Another viable route to this same end is the use of new developments in handwriting recognition algorithms. A select number of these approaches were assessed for viability and the One-Shot Learning technique was found to be most promising due to its hierarchical categorization system that is effective even with a limited dataset. Since there does not currently exist a database of handwritten hieroglyphs, the parameters of a beginning database were chosen. Subsequent work has focused on collecting samples for the database from a variety of different authors.

Current Status:

I am amassing a database on which to test the One-Shot Learning Technique