FIELD: physics, computer engineering.
SUBSTANCE: invention relates to an image processing. Disclosed is a method of processing a document image. The method includes a step of performing identification using a computing device comprising one or more processors, a plurality of images fragments within the document image. Further, the method comprises separating, using the computing device, the plurality of image fragments into a plurality of classes, wherein each class of the plurality of classes includes a subset of the plurality of image fragments which are substantially similar to each other, where determination of the degree of similarity of one image fragment to another is performed based on a given measure which determines the degree of similarity between said fragments.
EFFECT: high quality of source image fragments owing to filtering.
21 cl, 15 dwg
