چكيده به لاتين
Recognition of individuals based on their unique physiological or behavioral characteristics as biometric identification has been the focus of researchers in the past decade. A biometric detection system seeks to identify a person in a database. Among the existing biometric features, brain signals have emerged as a powerful feature due to the preservation of the unique nature of individuals, which makes forged or Imitating them impossible. Among the methods of receiving brain signals, Electroencephalography (EEG) is very popular and efficient as the easiest way to receive this signal.
Because brain activity can be affected by a person's mood, stress, and mental state, it is very difficult to record it under stress or intimidation, making it impossible to receive it in the above conditions that Can be mentioned as an advantage. To use EEG-based biometrics in real life requires a lot of research and different methods, the purpose of this dissertation is to evaluate this issue.
This dissertation uses the data of 109 people registered by 64 channels. Data are examined in two groups: open-closed and waiting. In the first step, the importance of preprocessing in identification is examined, which shows that preprocessing can be effective up to 7.5% in increasing the rate of performance. It also has a great effect on the stability of the final performance and prevents the scattering of results.
In the next steps, by extracting statistical, frequency, and wavelet features, the performance in two classification methods is investigated. Support vector machine (SVM) and multilayer artificial neural network are two classification methods that are examined.
The highest performance found in the waiting data is 97.43% by the neural network and in the open-closed data is 99.76% by the support vector machine. In this dissertation, the results of authentication using neural networks are very successful, especially on the waiting data, which is most similar to the application environment of authentication.
Keywords: Identification, Biometric, Electroencephalography, Support vector machine and multilayer artificial neural network.
