Font and Function Word Identification in Document Recognition

font would be used during recognition. This would reduce An algorithm is presented that identifies the predominant font in which the running text in an English language document the confusion caused by training on many fonts and would is printed. Frequent function words (such as the, of, and, a, effectively reduce the recognition problem to choosing the and to) are also recognized as part of the font identification. correct class from one font rather than from many fonts. Clusters of word images are generated from an input document One method for identifying the font in which a document and matched to a database of function words derived from is printed is to match the individual character images to fonts and document images. The font or document that matches the character images in a font database. The font of the best provides the identification of the predominant font and identified characters provides the desired information. A function words. This technique takes advantage of the fact that disadvantage of such a technique is that it only uses visual most machine-printed documents are prepared with a single information from isolated characters and is thus sensitive predominant font. Also, the repeated words in the document to noise that is commonly present in photocopies and facare utilized to overcome noise in the input. Advantages of this similes. technique include its use as a preprocessing step for a document recognition algorithm. Experimental results show high accuThe font identification algorithm presented in this paper racy is achieved on a database of original and degraded docudetects the predominant font in a given document, that is, ment images.  1996 Academic Press, Inc. the single font that is used in most documents to print