| 摘要 |
<p>PURPOSE: To identify an image based upon the distance data in plural directions with respect to a specific image. CONSTITUTION: Distance data on a specific image are generated in each of many start pixels in the image. Each start pixel is defined by a specific condition such as a side edge pixel and a pixel in a connection element. The distance data of each start pixel expresses a distance in each of many directions from the start pixel up to another pixel matched with another judgement condition for the image. In order to detect the distortion of lines in a text, e.g. distance data can indicate a distance from a start pixel on a side edge to the other side edge across a white pixel. In order to detect the type face of a main character, distance data can indicate a distance from a start pixel on the side edge of a connection element or in the connection element up to the other side edge across a black pixel. A division angle between adjacent directions can be sufficiently reduced to detect the feature of suitable size in a suitable distance, e.g. the feature of character size in an average character interval.</p> |
