FIELD: physics.
SUBSTANCE: invention relates to identification and detection of identity documents. In the method, at the first stage, an image of an identity document is supplied to the input; at the second stage, a machine-readable record (MRR) is detected on the image: a) applying computer vision and machine learning methods to search for a potential MRR on an image; b) if the MRR is not detected, the transition to the fifth step takes place; at the third stage, MRR is identified: a) applying optical character recognition to recognize and decode information in the MRR; b) if MRR does not comply with the standard, there is a transition to the fifth stage; c) extracting data on country of publisher and type of document from recognized data; fourth step involves searching for a document using a preselector: a) based on the coordinates of the found MRR, the document search area is set on the image, wherein in order to obtain a document search area, the area of the found MRR is increased as follows: upwards along the vertical the area is expanded by 4 times, if it is the MRR of two lines, and by 1.8 times, if from three lines, downwards by half the height of the MRR, and along the horizontal line, the zone is expanded by 20%; b) using the extracted country and document type information, using a preselector to search only among documents of the corresponding type and country; c) if a match is found, returning the result and terminating the algorithm; at the fifth stage, all types of documents and all countries are searched: a) if MRR is not detected or is not recognized, search is carried out among all available types of documents and all countries; b) if a match is found, returning a result; at the sixth stage, a result of document identification or a message stating that the document was not recognized is obtained.
EFFECT: high accuracy of identifying a document.
1 cl, 2 dwg
Title | Year | Author | Number |
---|---|---|---|
METHOD OF BINDING FIELDS OF CONDITIONALLY RIGID BUSINESS DOCUMENTS DURING THEIR RECOGNITION | 2024 |
|
RU2828605C1 |
DATA INPUT FROM IMAGES OF THE DOCUMENTS WITH FIXED STRUCTURE | 2014 |
|
RU2651144C2 |
METHOD FOR DETECTING FLUORESCENT FIBRES IN IMAGES OF IDENTITY DOCUMENTS IN THE UV SPECTRUM | 2022 |
|
RU2794418C1 |
METHOD OF RECOGNIZING PEOPLE ON VIDEO IMAGES | 2023 |
|
RU2820757C1 |
SYSTEM FOR DETECTING, SEARCHING, IDENTIFYING AND FIXING A FACE | 2019 |
|
RU2713615C1 |
METHOD FOR DETECTING FLUORESCENT ADHESIVE SPOTS IN IMAGES OF IDENTITY DOCUMENTS IN UV SPECTRUM | 2023 |
|
RU2814914C1 |
METHOD OF HANDWRITING SPECIMEN SIMILARITY ESTIMATION AND METHODS OF IDENTITY VERIFICATION AND HANDWRITING IDENTIFICATION USING THIS ESTIMATION METHOD | 2006 |
|
RU2340941C2 |
METHOD AND SYSTEM FOR EXTRACTING DATA FROM IMAGES OF SEMISTRUCTURED DOCUMENTS | 2015 |
|
RU2613846C2 |
METHOD OF INSERTING SECRET DIGITAL MESSAGE INTO PRINTED DOCUMENTS AND EXTRACTING SAID MESSAGE | 2010 |
|
RU2431192C1 |
METHOD OF ASSESSING THE VALIDITY OF A DOCUMENT USING OPTICAL RECOGNITION OF TEXT ON THE IMAGE OF A ROUND IMPRESSION OF A PRINT / STAMP ON A DIGITAL IMAGE OF A DOCUMENT | 2020 |
|
RU2750395C1 |
Authors
Dates
2025-04-07—Published
2024-06-27—Filed