Abstract: The present invention provides a method for generating derivative words including the steps of: creating a number of derivative grammar arrays; matching the inputting character information with the derivative grammar arrays and obtaining the match derivative grammar arrays; obtaining match words from the language database according to the condition arrays of the obtained derivative grammar arrays and the inputting character information; and generating derivative words by adding the suffix alphabetic character sets of the obtained derivative grammar arrays to the ends of the words. In accordance with the established grammar rules, the words in the language database can be converted to derivative words and the derivative words do not need to be stored in the language database. Therefore, the storage space of the language database can be remarkably reduced. The present invention also provides a system for generating derivative words.

Abstract: The present invention discloses a method for recognizing a handwritten character, which includes the following steps of: obtaining a coarse classification template and a fine classification template; receiving a handwritten character input signal from a user, gathering a discrete coordinate sequence of trajectory points of the inputted character, and pre-processing the discrete coordinate sequence; extracting eigenvalues and calculating a multi-dimensional eigenvector of the inputted character; matching the inputted character with the coarse classification template to select a plurality of the most similar candidate character classes; and matching the eigen-transformed inputted character with sample centers of the candidate character classes selected from the fine classification template, and determining the most similar character classes among the candidate character classes. The present invention further discloses a system for recognizing a handwritten character.

Abstract: The present invention discloses a method for recognizing a handwritten character, which includes the following steps of: obtaining a coarse classification template and a fine classification template; receiving a handwritten character input signal from a user, gathering a discrete coordinate sequence of trajectory points of the inputted character, and pre-processing the discrete coordinate sequence; extracting eigenvalues and calculating a multi-dimensional eigenvector of the inputted character; matching the inputted character with the coarse classification template to select a plurality of the most similar candidate character classes; and matching the eigen-transformed inputted character with sample centers of the candidate character classes selected from the fine classification template, and determining the most similar character classes among the candidate character classes. The present invention further discloses a system for recognizing a handwritten character.