| 摘要 |
PROBLEM TO BE SOLVED: To surely decompose an integrated original into each image even though there is a specific image, such as handwriting, near a boundary region of the integrated original. SOLUTION: An original obtained by integrating a plurality of images is read, and a layout analyzing part 30 analyzes image regions and character regions and obtains an image for performing character recognition. A character recognizing part 32 performs character recognition about an image for performing character recognition to obtain a character recognition image whose character recognition is successful, and an integration form is determined by comparing the character recognition image with a predetermined pattern for integration form determination. In addition, an attribute data extracting part 36 extracts a non-character image existing in the boundary region of the image for performing character recognition, and an attribute data attaching part 42 attaches attribute data representing the non-character image as a page attribute. COPYRIGHT: (C)2009,JPO&INPIT
|
