METHOD AND DEVICE FOR CLASSIFICATION OF IMAGES OF PRINTED COPIES OF DOCUMENTS AND SORTING SYSTEM OF PRINTED COPIES OF DOCUMENTS

摘要

FIELD: printing industry.;SUBSTANCE: images of a printed copy of the document are entered into a number of means of extracting image descriptors. Image descriptors are extracted. Each image descriptor describes the characteristics of the image in the printed copy of the document. After that, a pularity of trained classifiers is used to evaluate the probability of the document printed copy image classes based on the extracted image descriptors. The most likely image class of the document printed copy is determined using a trained meta-classifier using the estimated class probabilities. An image of the document printed copy and the most likely class of the document printed copy are then entered into the assigning tool. The most likely class determined by the trained meta-classifier is assigned by the assigning tool to the image of the document printed copy to obtain a classified image of the document printed copy.;EFFECT: providing the possibility of automatic image classification of the document printed copies.;33 cl, 8 dwg