Mood recognition based on natural scenes for locomotive driver

The locomotive drivers keep calm mood is the first condition in driving high-speed rail safely. Since different scenarios during their journeys tend to evoke a wide range of moods, how to infer the driver's emotional state as well as cause a warning is the primary means to reduce mood swings. This paper proposes a mood recognition method. Specially, the 1 / ƒ fluctuation is the main measurement of Electroencephalography (EEG) which is a useful tool to analyze human emotional states. The HSV space has strong association with mood model. Therefore, this paper calculates the slopes of the power spectra on HSV as the affective characteristics of natural scenes. And then the K-nearest neighbor (KNN) classifier is used to differentiate the various mood categories based on the affective characteristics. We show results for the proposed approach on the International Affective Picture System (IAPS), a standard mood evoking image set in psychology. The promising results demonstrate that the effectiveness of affective representation to model the mood content of natural scenes.