METHOD FOR RECOGNIZING HEAD-WRITTEN CHARACTER BY FUZZY INFERENCE

摘要

PURPOSE:To shorten the processing time necessary for recognition by executing the Fourier transformation corresponding to X and Y moving quantities in units of strokes for the normalized hand-written character, obtaining intensity for the frequency and further, making a fuzzy inference. CONSTITUTION:Characters hand-written to a media graph 1 is inputted to a personal computer 2 as coordinate point sequence data corresponding to the X coordinates and the Y coordinates at the process where the character is hand-written and the coordinate point sequence of respective strokes of the hand-written character is transferred to an input data normalizing part 3 of a computer 2 in accordance with the input sequence. The normalizing part 3 normalizes the size and writing speed of the character, in a Fourier transforming part 4, the Foulier transformation corresponding to the X and Y moving quantities is performed in units of strokes and the intensity of the frequency is obtained. The intensity corresponding to the respective frequencies is sent to a fuzzy part 5, and the data, put to a fuzzy algorithm is fuzzy-infered by a fuzzy inference part 8 using the pattern data stored in a standard pattern part 6. Thus, the processing time of the character recognition can be shortened.