YOSHIDA HisashiDepartment of Computational Systems Biology Professor/Manager |
Even when the eye gazes at a certain point of view, the eye moves involuntarily and constantly, which is called fixation eye movement. The fixation eye movement is mainly composed of three components, microsaccade, drift, and tremor. In recent years, there have been many reports showing that the fixation eye motion is under the influence of cognitive mechanisms such as attention. Therefore it is an important index for knowing human cognitive mechanisms. In this research, we propose a new method that tracks the fixation eye movement by using a state space model. The model can separate the fixation eye movement into three components, microsaccade, drift, and tremor. As a result, it was possible to accurately track the fixation eye movement even including the overshoot and undershoot of the microsaccade. In addition, we proposed a new method for microsaccade detection using the median prediction and the Bayesian prediction interval of the fixation eye motion tracking. Compared with conventional methods that detect micro saccades using smoothing differentiation and thresholds, the proposed method can detect microsaccades with high accuracy with taking into account overshoots and undershoots that appear in micro saccades, and the detection rate was also high.