Palaeography and Image Processing: Some Solutions and Problems

This paper considers the application of image-processing and data-mining to the analysis of scribal hands. The work of forensic document analysts on feature-extraction is considered, particularly the algorithms developed for automatic handwriting-recognition by Srihari, and by Bulacu and Schomaker. Automatic clustering is also considered using the AutoClass package. Preliminary results of the author’s own experiments with these approaches are presented, and some of the obstacles are outlined which must be overcome before a practical system can be developed for the automatic identification of medieval scribes.