Determination of the 25th frame effect stored in the videos with the EEG signals

The videos that come across every part of our life are a set of images resulting from sequential sequencing of a series of image files. The number of pictures displayed every 1 second of the video is called Frame Per Second (Fps). With 24 consecutive picture frame, one second of the video is formed, and the images that are hidden in these frames and called the 25th frame effect are seen difficult by the human eye in the video. The aim of this thesis is to determine whether the brain will detect hidden pictures (25th frame) in videos that are thought to affect people's subconcious. For this aim, 6 different videos with animal, plant and nature theme which each of them include 25th frame effect, and raw videos are prepared in order to be watched by participants. These videos were shown to 50 participants. While viewing the videos, wireless heading Emotiv EPOC+ device which record 16 channel Elektroansefalogram (EEG) signal, 2 of which are reference, was used with a non invasive way on the scalp of the participants. The recorded EEG signals were processed by Fast Fourier Transform (FFT) and Hilbert Huang Transform (HHT) filtering methods. Statistical features extraction algorithms (maximum and minimum difference, mean, median, standard deviation, power, variance, energy, kurtosis, skewness, interquartile range) was applied to the obtained datas. In order to compare the signals generated while watching the raw picture of the videos and the videos that were recorded with hidden pictures, one of the K classification methods, nearest neighbor (KNN) classifier and the linear classifier Naive Bayes (NB) was used. Training was provided by applying K-part cross validation method. The classification was performed using three different methods: raw, HHT-applied and FFT-applied of EEG signals. Achievement rates which were obtained are very high.