| 摘要 |
PURPOSE: To efficiently decide a character kind in an inputted document image and improve the process precision of a character recognizing process by deciding the character kind of a line of interest according to line height, the number of characters, and the width of a tentative character detected by a detecting means. CONSTITUTION: A bit map image inputted through an image scanner 6 is divided into blocks first. When a block to be processed at present includes characters, the data in the block are projected on a Y axis and the lines where the characters are present is extracted. Then respective lines are projected on an X axis and the characters are segmented. It is judged whether or not the longitudinal/lateral ratio of the character width and height of each character is larger than a previously set threshold value. When the judgement result is affirmative, it is judged that the line of interest consists of Japanese characters and the character recognizing process is instructed to refer to a Japanese dictionary or recognize the characters by referring to it preferentially. When the judgement result is inaffirmative, on the other hand, it can be judged that the line of interest consists of English characters, so the recognizing process is instructed to refer to an English recognition dictionary or perform the processing by referring to it preferentially. |
