METHOD AND SYSTEM FOR SEARCHING GRAPHIC IMAGES Russian patent published in 2023 - IPC G06V30/18 G06F16/30 

Abstract RU 2807639 C1

FIELD: computer technology.

SUBSTANCE: invention relates to the processing of digital data to search for graphic information. The claimed invention is carried out by implementing a computer-implemented method for searching graphic images, performed using a processor and containing stages in which: the input image is processed using OpenCV and Pillow library algorithms, with the help of which the image is converted to grayscale and noise is removed, wherein when removing noise, high-contrast pixels of the input image against a background of the same colour are removed; using a machine learning model based on an artificial neural network (ANN), trained on graphic images and containing a database of their vector representations, the input image is processed, during which the image is converted into a pixel matrix, from which the key parameters of the image and their location in input image are determined, and a vector representation of the image using a maximization algorithm is formed; a search index for the input image by comparing the proximity of the vector representation of the input image with the vector representations in the ANN database is generated, and the comparison is performed using an approximate nearest neighbour search algorithm based on hierarchical graphs that allow clustering the resulting vectors in such a way that within one a cluster contains vectors that differ minimally from each other in the maximum number of elements; determining and providing at least one image from the database based on a comparison of the vector representations against said search index.

EFFECT: increasing the accuracy and speed of searching for graphic information.

11 cl, 7 dwg

Similar patents RU2807639C1

Title Year Author Number
DEVICE FOR SEMANTIC CLASSIFICATION AND SEARCH IN ARCHIVES OF DIGITIZED FILM MATERIALS 2016
  • Podlesnyj Sergej Yurevich
  • Kucherenko Aleksej Valentinovich
RU2628192C2
SYSTEM AND METHOD FOR AUGMENTATION OF THE TRAINING SAMPLE FOR MACHINE LEARNING ALGORITHMS 2020
  • Shavrina Tatyana Olegovna
RU2758683C2
METHOD AND SYSTEM FOR CHECKING MEDIA CONTENT 2022
  • Gorb Roman Viktorovich
  • Yudin Sergej Mikhajlovich
  • Zobnin Aleksej Igorevich
  • Oreshin Pavel Evgenevich
RU2815896C2
METHOD OF CREATING MODEL FOR ANALYSING DIALOGUES BASED ON ARTIFICIAL INTELLIGENCE FOR PROCESSING USER REQUESTS AND SYSTEM USING SUCH MODEL 2019
  • Antyukhov Denis Olegovich
  • Pugachev Leonid Petrovich
RU2730449C2
METHOD AND SYSTEM FOR STORING MULTIPLE DOCUMENTS 2018
  • Shlyunkin Aleksey Valerievich
  • Uvarov Nikita Denisovich
  • Boymel Aleksandr Alekseevich
  • Gotmanov Aleksandr Nikolaevich
RU2744028C2
METHOD AND SYSTEM FOR TRAINING CHATBOT SYSTEM 2023
  • Zinov Nikolaj Aleksandrovich
  • Korenev Artem Arkadevich
RU2820264C1
METHOD AND SERVER FOR DETERMINING TRAINING SET FOR MACHINE LEARNING ALGORITHM (MLA) TRAINING 2020
  • Dorogush Anna Veronika Yurevna
  • Alipov Vyacheslav Vyacheslavovich
  • Kruchinin Dmitriy Andreevich
  • Oganesyan Dmitry Alekseevich
RU2817726C2
METHOD AND SYSTEM FOR EXPANDING SEARCH QUERIES IN ORDER TO RANK SEARCH RESULTS 2018
  • Gotmanov Aleksandr Nikolaevich
  • Grechnikov Yevgeny Aleksandrovich
  • Safronov Aleksandr Valerievich
RU2720905C2
METHOD OF TRAINED RECURRENT NEURAL NETWORK DEBUGGING 2019
  • Zharov Yaroslav Maksimovich
  • Korzhenkov Denis Mikhajlovich
RU2715024C1
METHOD AND SYSTEM OF SELECTION FOR RANKING SEARCH RESULTS USING MACHINE LEARNING ALGORITHM 2018
  • Duzhik Valery Ivanovich
  • Drozdovsky Andrey Dmitrievich
  • Naydin Oleg Pavlovich
RU2731658C2

RU 2 807 639 C1

Authors

Shulga Sergej Aleksandrovich

Dates

2023-11-20Published

2022-07-22Filed