Author

Ji Won Chung

Publication Date

2018-05-14

Document Type

Honors Project

Degree Name

Bachelor of Arts

Department

Computer Science

Advisors

Nicholas R. Howe and Gwen Spencer

Keywords

Word spotting, Graph-based matching, Inkball models, Part-structured inkball models, Segmentation-free, Handwriting recognition, Document analysis, Word recognition-Mathematical models, Statistical matching, Paleography

Abstract

Handwritten historical manuscripts traditionally have been manually transcribed for the purpose of preservation and dissemination. Part-structured Inkball models provide a potentially segmentation-free word spotting method that automates query searches in large collections of texts. However, the existing technique considers the graphical structure of words only to a limited extent. This thesis proposes a new method to measure the similarity of two inkball models. We first propose a bidirectional match between two graphs. Then we introduce two new measures to capture many-to-one matches of nodes and the structural differences between graphs. Tested on the standard George Washington 20 dataset, the method shows modest improvement in comparison to the original technique.

Rights

2018 Ji Won Chung. Access limited to the Smith College community and other researchers while on campus. Smith College community members also may access from off-campus using a Smith College log-in. Other off-campus researchers may request a copy through Interlibrary Loan for personal use.

Language

English

Comments

vi, 43 pages : color illustrations. Includes bibliographical references (pages 42-43)

Share

COinS