EEG Based Emotion Recognition using Uniform LBP

Electroencephalography is a commonly used technique for capturing brain signals in brain-computer interface systems. Multiple metrics can be used to detect emotions, including subjective self-reports, autonomic and neurophysiological data, or their combinations. Present EEG-based emotion recognition systems have lower performance, higher computational complexity, or both. This paper proposes the use of a one-dimensional local binary pattern with uniform sequence for the analysis of EEG signals obtained from DEAP dataset. It captures local spatial features, reducing the amount of data needed to represent the signal while preserving critical textural information. The suggested approach demonstrates improved effectiveness compared to the latest algorithms in recognizing human emotions. In the analysis, SVM and KNN classifiers shows superior performance compared to other classifiers, achieving higher accuracy and more reliable results. A subject-specific and channel-wise analysis is also performed in this study.