| 摘要 |
PROBLEM TO BE SOLVED: To extract and recognize only the written characters serving as the data to be recognized by comparing a document image (1st document image) obtained before the characters are written with a document image (2nd document image) obtained after the characters are written, erasing the previously printed character images out of the 2nd document image and extracting and recognizing only the written characters. SOLUTION: A ruled line erasing part 10 projects the information on the ruled lines shown on a master image that is registered on a data image and erases these ruled line images. A coupling component erasing part 14 replaces the coupling components which are decided equal to each other with the while pixels to erase the images of preprinted characters. A character image extracting part 15 extracts the coupling components in a recognizing object area of the data image where the images of ruled lines and the preprinted characters are erased and also extracts the character images written as the data to store them in a character image memory 16. A character recognizing part 17 recognizes the extracted character images, converts these recognized images into the character codes and outputs these codes.
|
