Arousal estimation method based on optical flow analysis for motion graphics

Abstract

Motion graphics is a form of visual expression characterized by non-narrative, non-figurative based visuals that change over time. Due to the expansion of its application areas, a consideration of the affective quality of motion graphics is growing more important. This paper proposes an arousal estimation method that uses optical flow analysis as an affective quality assessment method for motion graphics. The primary objective is to verify two indexes: the total flow amount and the average magnitude of motion vectors. According to correlation analyses between each index and the arousal factor scores of video stimulus derived from an impression test with human participants, the correlation between the average magnitude of motion vectors and the arousal factor scores is significant. An analysis of the distribution of displacements showed that about three pixels per 33.3 msec is the border value: a higher ratio of movement slower than that border reduces the arousal level, and faster movement raises the level.