METHODS OF IDENTIFICATION OF A HUMAN SIGNATURE
Authors:
1.
KNRTU named after A.N. Tupolev
2.
Kazan National Research Technical University named after A.N. Tupolev
3.
KNITU im. A.N. Tupoleva (KAI)
Type:
Article
Pages:
from 147
to 152
Status:
In work
Language:
Russian
Keywords:
SIGNATURES, IDENTIFICATION, TEXTURE, CLASSIFICATION, PROGRAMMING
Abstract (English):
The article explores methods and approaches to identify signatures of specific people. The main concepts and principles related to signature processing and analysis, including image preprocessing, feature extraction and signature classification are considered. It is indicated that a signature is a biometric feature that a person can use without the need to remember passwords or use additional devices. Various identification methods help solve the problem, for example, methods that take into account the shape, texture, and structure of the signature, speed and acceleration characterize the dynamics of signatures. In some cases, machine-learning methods are appropriate. Each of these approaches has its own advantages and limitations. There are the following signature identification methods: Template Matching, Feature Extraction and Neural Network-based Method. Their characteristics, as well as the advantages and disadvantages of these methods are given. The authors used a mathematical model to extract features: Bayesian classification, which uses probabilistic models to determine the class of an object based on its features. For software implementation, we used the Python programming language and OpenCV, an open-source computer vision library. These means help to process the signature in the following sequence: image loading; image processing; signature segmentation and image saving. A data set containing signature samples from different people provided verification the results. During the program implementation, images are pre-processed using feature extraction and classifier training based on the Bayesian approach.
The article explores methods and approaches to identify signatures of specific people. The main concepts and principles related to signature processing and analysis, including image preprocessing, feature extraction and signature classification are considered. It is indicated that a signature is a biometric feature that a person can use without the need to remember passwords or use additional devices. Various identification methods help solve the problem, for example, methods that take into account the shape, texture, and structure of the signature, speed and acceleration characterize the dynamics of signatures. In some cases, machine-learning methods are appropriate. Each of these approaches has its own advantages and limitations. There are the following signature identification methods: Template Matching, Feature Extraction and Neural Network-based Method. Their characteristics, as well as the advantages and disadvantages of these methods are given. The authors used a mathematical model to extract features: Bayesian classification, which uses probabilistic models to determine the class of an object based on its features. For software implementation, we used the Python programming language and OpenCV, an open-source computer vision library. These means help to process the signature in the following sequence: image loading; image processing; signature segmentation and image saving. A data set containing signature samples from different people provided verification the results. During the program implementation, images are pre-processed using feature extraction and classifier training based on the Bayesian approach.
Keywords:
SIGNATURES, IDENTIFICATION, TEXTURE, CLASSIFICATION, PROGRAMMING
SIGNATURES, IDENTIFICATION, TEXTURE, CLASSIFICATION, PROGRAMMING
