摘要 |
PURPOSE:To heighten serviceability of pattern recognition by utilizing the content meant by input pattern from partial recognition information to input pattern and recognizing the content of whole input pattern presuming the content of uncuttable part or unrecognizable part in the input pattern. CONSTITUTION:An inputting section 1 cuts an inputted picture and detects individual character and picture which is the unit of object of recognition. When it is discriminated that plural characters are included in a character picture from the length of an area segmented for instance, as a character picture, an uncuttable mark (*) is given to the area as the first pointer. Further, an unrecognizable mark (?) is given to a character picture decided as unrecognizable in the unit of object of recognition to which the first pointer is not given as the second pointer. An afterprocessing section 3 makes the result of recognition of each character picture obtained by a recognizing section 2 and relative position information of the character picture in inputted character string a knowledge information, and makes the content of object of recognition another knowledge information, and presumes the content (information) of the character information to which pointers are given using these knowledge information.
|