Indexing for a Digital Library of George Washington’s Manuscripts: A Study of Word Matching Techniques
T.M. Rath, Sean Kane, Abderrafih Lehman, Eric Partridge, Raghaven Manmatha · ScholarWorks@UMassAmherst (University of Massachusetts Amherst) · 2002
In a multimedia world, one would like electronic access to all kinds of information. But a lot of important information still only exists on paper and it is a challenge to e#ciently access or navigate this information even if it is scanned in. The previously proposed "word spotting" idea is an approach for accessing and navigating a collection of handwritten documents available as images using an index automatically generated by matching words as pictures. The most di#cult task in solving this problem is the matching of word images. The quality of the aged documents and the variations in handwriting make this a challenging problem. Here we present a number of word matching techniques along with new normalization methods that are crucial for their success. E#cient pruning techniques, which quickly reduce the set of possible matches for a given word, are also discussed. Our results show that the best of the discussed matching algorithms achieves an average precision of 73% for documents of reasonable quality.