| 摘要 |
A method and apparatus for correctly cutting an unidentified read composite character image into separate and discrete character images suitable for use in recognizing characters in which the images have been wrongly joined to each other to form a composite image when read from an original. For the purpose of correctly determining the position where the composite image is to be cut into separate character images, the apparatus has a configuration tracing device for detecting coordinate information concerning the configuration of an image of interest in input image information from an original. A distance computing device computes, on the basis of the coordinate information obtained by the configuration tracing device, the distances between the ends of the configuration traced by the configuration tracing device as measured in the direction perpendicular to the train of characters, for successive points along a line parallel to the train of characters. A separation position determining device determines, on the basis of the distance information obtained by the distance computing device, a position where the image is to be cut into separate images.
|
