Browsing by Author "Melek, Mesut"

Now showing 1 - 1 of 1

Detailed Channel/Lob Analysis of a 2d and 3d Hybrid Video Using Eeg Signal
(Konya Technical University, 2021) Manshouri, Negin; Melek, Mesut; Kayikçioğlu, Temel
It is important to evaluate sudden transition and steady-states in video analysis consisting of 2 dimensional (2D) and 3D tracks, regarding the human eye structure. In this study, the power spectrum density (PSD) of the human brain signals was taken into consideration for analysis as a result of a 2D and 3D hybrid video watching. We claimed in our previous studies that people lose their depth perception when they get tired. In this study, a single stream anaglyph video consisting of random 2D and 3D tracks is provided. In 2D and 3D hybrid video study, PSD based on short-time Fourier transform (STFT) and visual representation called spectrogram were used in brain signal analysis. After all EEG frequency bands have been tested, the delta band has been chosen as the dominant band, taking into account the difference of PSD in the comparison of 2D and 3D parts in the spectrogram chart. Extracted features from this band were classified by two popular classifiers. These are support vector machine (SVM) and Linear discriminant analysis (LDA) algorithms. Consequently, the frontal and temporal lobes show better results in the classification of 2D and 3D transitions than other lobes. As a result of statistical functions and Hjorth parameters used as feature extraction methods, classification success for SVM and LDA algorithms was computed as 68 % and 79 %, respectively.