摘要 |
A peripheral distribution of filled pixels in a document image is calculated by projecting the filled pixels in an X-axis or a Y-axis direction. A bottom part in the peripheral distribution is detected. The document image is divided into a plurality of primary image regions in accordance with a dividing line intersecting the bottom part in the X-axis or Y-axis direction, so that the document image is classified into text regions, drawing regions and picture regions. Thus, the text regions can be extracted automatically from the document image without requiring a specific manual operation for extracting text regions by an operator.
|