摘要 |
PURPOSE:To understand the meaning of a document whose form is unknown by scanning longitudinally and laterally an input picture and comparing the mean value of longitudinal and lateral inter-character interval. CONSTITUTION:The input picture is scanned longitudinally and laterally to obtain the histogram of picture elements forming a character section, the mean value of the inter-character interval in longitudinal and lateral directions obtained by the histogram is compared to extract the direction of line of the document simply and recognized candidate characters are edited. that is, the rectangular R of the segmented recognition object character inputted sequentially to the recognition section 5, as to each picture element of the segmented character, whether or not M-set of elements or over are connected including the noted picture element in the direction of arrow is checked to set a direction code. The connection of each picture element is checked to extract the stroke at each direction code and the number, position and length of the stroke (characteristic amount) are extracted. The extracted characteristic amount are collated with that of the character registered in a dictionary 6 and the most similar character is selected as a recognition candidate character.
|