METHOD AND SYSTEM FOR RECOGNIZING CONTEXTUAL CHARACTER OR PHONEME

摘要

PURPOSE: To improve the recognition of an erroneously segmented text by using character candidate string recognition. CONSTITUTION: A candidate discriminator 102 receives information related to a text to be recognized and provides character candidate identifiers and a string recognizer 104 recognizes the string of the candidate identifrers and provides the string of characters corresponding to the text based on the recognized string. That 15, the string recognizer 104 is provided with plural recongition devices 110 and 120 respectively provided with processors 112 and 122 and sub set data structures 114 and 124 and measures whether or not the received string of the candidate identifiers is present in the sub set of the respective candidate identifier strings. Thus, erroneously segmented character candidates inside a sample text are led to correctly recognized characters.