Size of the character set with which it has to deal is the main area of concern during development of any OCR. The presence of larger number of character (theoretically 46656) in the middle zone of the Devanagari Text makes the development of OCR for Devanagari complex. Hence every effort is required to break this larger character set into small manageable chunks so that recognizer has only to deal with a small subset of symbols. The identification of various types of structural properties which are invariant to fonts and sizes reduces the bigger character set of Devanagari Text into small manageable chunks. This set can be reduced further in each class if more structural properties are identified. This results in development of a recognizer with increased accuracy and better response time. One such structural property is the number of cavities present in the left surface of the character. This structural property further reduces the character set identified by the various other structural properties viz. bar type, number of shirorekha touching and water bodies.
