|
SCIENTIFIC COUNCIL AWARDING SCIENTIFIC DEGREES DSc.13/30.12.2021.T.142.01 AT RESEARCH INSTITUTE FOR
|
| bet | 34/36 | | Sana | 14.05.2024 | | Hajmi | 1,65 Mb. | | #233056 |
Bog'liq amaliyot kundalik SCIENTIFIC COUNCIL AWARDING SCIENTIFIC DEGREES DSc.13/30.12.2021.T.142.01 AT RESEARCH INSTITUTE FOR
DEVELOPMENT OF DIGITAL TECHNOLOGIES AND ARTIFICIAL INTELLIGENCE
RESEARCH INSTITUTE FOR DEVELOPMENT OF DIGITAL TECHNOLOGIES AND ARTIFICIAL INTELLIGENCE
ABDUKADIROV BAKHTIYOR ABDUVAXITOVICH
ALGORITHMS FOR DETECTING FALSE INPUT DATA IN BIOMETRIC PERSONAL IDENTIFICATION SYSTEMS
05.01.03 – Theoretical basis of computer science
DISSERTATION ABSTRACT OF THE DOCTOR OF PHILOSOPHY (PhD) ON TECHNICAL SCIENCES
Tashkent-2022
39
40
INTRODUCTION (abstract of PhD dissertation)
The aim of the research work is to develop algorithms for detecting false attacks on biometric personal identification systems based on face images, as well as to create a software package based on them.
The object of the research work is the systems for identification and authentication.
The scientific novelty of the research work is as follows:
an algorithm for geometric normalization of the face relative to the horizontal axis based on special points of facial images, which is part of the biometric system, has been developed;
an algorithm for detecting false input data in biometric personal identification systems based on texture analysis by constructing image descriptors and histograms has been developed;
a method for determining the liveliness of the face, taking into account moirés and blurs in the video image has been developed;
an algorithm for identifying the authenticity of data included in the biometric identification system based on face images by deep learning of neural networks has been developed.
Implementation of the research results. Based on software created on the basis of existing and proposed models, methods and algorithms for determining the authenticity of data entering biometric identification systems based on facial images:
a software package developed on the basis of algorithms for normalizing incoming facial images and determining the authenticity of information entered into the biometric identification system has been introduced in the Main Department for the Development and Digitalization of Logistics of the Joint Stock Company «Uzbekiston Temir Yullari» to solve the problems of two-factor authentication when controlling the entry and exit of workers to special premises (Reference of the Joint-stock company
«Uzbekiston temir yullari» No. 03/12-114 dated August 31, 2022). As a result, it was possible to increase the security of entry and exit to special premises;
a software package developed on the basis of deep learning algorithms for normalizing incoming facial images and determining the authenticity of incoming data was introduced to control access to protected facilities in the Department of Military Security of the Joint-Stock Company «Uzbekistan Temir Yullari» (Reference of the Joint-Stock Company «Uzbekiston Temir Yullari» No. 03/12-114 dated August 31, 2022); As a result, it was possible to reduce the time for entry-exit control by an average of 10%;
a software package developed on the basis of a deep learning algorithm for determining the authenticity of data entering biometric identification systems from face images, and a method for determining the liveliness of a face from a sequence of video frames, has been implemented in the Department of Technical and Technological Control, the Kamchik Tunnel Coordination and Management Service of the Joint Stock Company society «Uzbekiston temir yullari», introduced two-factor authentication for exit control (Reference of the Joint Stock Company «Uzbekiston temir yullari»
No. 03 / 12-114 dated August 31, 2022). As a result, it was possible to increase the safe throughput of entry and exit by 1.1 times.
|
| |
