Please use this identifier to cite or link to this item: https://elar.rsvpu.ru/handle/123456789/28013
Title: The technology of «computer vision» in the question of visual identification of a person
Authors: Kuznetsov, R. A.
Ushakov, M. A.
Maschenko, M. V.
Volkova, E. A.
Issue Date: 2017
Publisher: National Research Nuclear University
Abstract: Describes the results of work in the field of visual identification of the user using the application system. The detailed description of the structure of the system, its interface and painted the algorithm of the system, providing visual identification of the human face. In the basis of the presented algorithm in system identification LBPH, which is currently the most advanced in the field of two-dimensional identification of a facial image. Presented designed the class diagram of the system developed. Described algorithms of the system operation in the storage mode of the object and in the identification mode. Based on the developed class hierarchy and algorithms of the system was written in code in the programming language C ++. As a result, presented an application that creates a collection of training samples from a set of visual images of the object being initialized, and also produces the training of the machine learning algorithm and performs the visual identification of a person based on the vector model obtained by machine learning.
Keywords: APPLICATION SOFTWARE
COMPUTER VISION
DATA ANALYSIS
IDENTIFICATION
VISUALIZATION
APPLICATION PROGRAMS
ARTIFICIAL INTELLIGENCE
C++ (PROGRAMMING LANGUAGE)
COMPUTER PROGRAMMING
DATA REDUCTION
DATA VISUALIZATION
DIGITAL STORAGE
FLOW VISUALIZATION
IDENTIFICATION (CONTROL SYSTEMS)
IMAGE PROCESSING
LEARNING ALGORITHMS
LEARNING SYSTEMS
APPLICATION SYSTEMS
CLASS DIAGRAMS
CLASS HIERARCHIES
FACIAL IMAGES
SYSTEM OPERATION
TRAINING SAMPLE
VECTOR MODELS
VISUAL IDENTIFICATION
COMPUTER VISION
ISSN: 2079-3537
SCOPUS: 85018781790
Appears in Collections:Научные публикации, проиндексированные в SCOPUS и WoS

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.