| 摘要 |
FIELD: image recognition technologies. ^ SUBSTANCE: method includes dividing images on areas, separating text and non-text areas, recognizing objects, prior assignment of at least one graphic construction made of least units of image as typical element for compiling recognized object, prior comparing of image of said object in form of a set of typical elements of same type to their mutual space-parametric relations. Method also includes searching and identification of image of typical element, selection of found typical element for checking whether it belongs to recognized object, generation and checking of hypothesis about recognized object on basis of image from set of selected typical elements as a whole with consideration of their mutual space-parametric relations. ^ EFFECT: higher precision, higher resistance to interference. ^ 2 cl, 3 dwg |
