LAUSR

With the rapid development of eye tracking technology, eye movements have become more and more important in human–computer interaction. Generally, eye movements are classified into fixation, saccade, and smooth pursuit. Since the eye movements are natural and fast, contain important cues for human cognitive state and visual attention, the eye movement behaviors are difficult to detect and classify. In this paper, the novel eye-movement data filtering and eye-movement classification algorithm are proposed. The nonlinear wavelet threshold denoising method was used to smooth the eye-movement data and detect saccades in the presence of smooth pursuit movements, according to different eye-movement behaviors related to the different characteristics of wavelet detail coefficients. Experiments were conducted to compare the eye-movement signal analyzing algorithm based on wavelet with other algorithms. The results showed that the eye-movement data filtering algorithm based on wavelet performed better than the other eye-movement filters. Moreover, the classification algorithm based on wavelet can classify different eye-movement behaviors more accurately. Then, we used an eye tracking technology to record and analyze the user’s eye movement during the test, so as to get the user’s psychological and cognitive state.