| 摘要 |
PURPOSE:To obtan a high recognition factor for both handwritten characters and type characters by referring to the value obtained by averaging the square sums of all feature variable constants in order to discriminate the handwritten and type characters from each other. CONSTITUTION:The variances DELTAx, DELTAy, DELTAz... from a dictionary variable constant Pd to each of feature variable constants (x, y, z...) and a variance DELTAR between the Pd and the data R obtained by integrating all said feature variable constants are compared with the threshold value respectively. Then the threshold levels DELTAxB, DELTAyB, DELTAzB... and DELTARB of handwritten characters are set in an initial state. A dictionary is retrieved for each feature variable constnt extracted from characters to be recognized to extract the differences smaller than the prescribed value by means said threshold values of handwritten characters and to define them as the first candidate characters. Then these characters are checked to obtain a distance DELTAR from the Pd for each feature variable constant. This distance DELTAR is compared with the threshold value DELTARB to obtain the distance DELTAR, and the sum total of DELTAR is obtained by a register AD. This obtained sum is divided by the character number to obtain the mean value of distance DELTAR.
|
