Unsupervised classification for ancient manuscript analysis

Lu, Karen, Psarrou, Alexandra, Licata, Aaron, Konstantinou, Vassilis and Kokla, Vassiliki (2008) Unsupervised classification for ancient manuscript analysis. In: Proceedings of the International MultiConference of Engineers and Computer Scientists (IMECS'08), 19-20th March 2008, Hong-Kong. International Association of Engineers, pp. 348-353. ISBN 9789889867188

Full text not available from this repository.


Ancient manuscript analysis is to aid historian to classify, annotate and judge the authenticity of larger collections of ancient manuscripts. Previous method was to examine the composition of manuscript such as paper support or inks by destructive sampling and chemical analysis. The aim of this paper is to present an image-based non-destructive and non-invasive method to analyze ancient manuscripts. A new unsupervised classification algorithm was designed to distinguish different type of paper support without any priori knowledge. The advantage of this classifier comparing with traditional methods such as K-means is that no pre-defined class number is needed. This is critical important in this application because no any priori knowledge about these manuscript was available before classifying. We presented the multi-model based framework of unsupervised classification, so that different statistical models can be applied to the same flow chart for different applications. This classifier enables us to study the statistical properties of these manuscripts, which is very hard to characterize due to the decay materials and bad resolutions. We evaluated this unsupervised classification performance by a specially developed algorithm. Experiment results show the potential of our unsupervised classification method in ancient manuscript analysis.

Item Type: Book Section
Additional Information: Best paper award
Uncontrolled Keywords: Texture Analysis, Ancient Manuscript Analysis, Unsupervised Classification, K-means, Clustering
Subjects: University of Westminster > Science and Technology > Electronics and Computer Science, School of (No longer in use)
Depositing User: Miss Nina Watts
Date Deposited: 20 Jan 2009 15:23
Last Modified: 14 Oct 2009 11:29
URI: http://westminsterresearch.wmin.ac.uk/id/eprint/5656

Actions (login required)

Edit Item (Repository staff only) Edit Item (Repository staff only)