| 摘要 |
An image processing method operates on a colored digitised image in a computer ( 2 ) to derive the image in abstract computer representation. Color is used to pre-classify symbols and other features of the digital image in order to simplify subsequent character recognition steps. By preparing the image in accordance with rules such that a unique color is used for all features of a defined set, and by storing those features in the computer in computer representation in corresponding sets, once a match has been made between a graphic feature and a stored feature, then the color of that graphic feature is made unique to the set of stored features containing the matched feature. The remaining graphic features of that color then need to be compared only with the stored features in the same set, rather than with all the stored features (of all the sets), thus significantly reducing the character recognition processing that is required to effect matching.
|
