Time-Sequential Analysis of Eye Movements Induced by Features of Given Images

Abstract

A great number of uncertainities are included in behavior of the eyes. So far, a factor of causing the saccadic eye movements has not explicitly been found. To elucidate the possible factor, visual experiments are performed : different textures in a large display are presented to the subjects with the eye-tracker for 20 seconds, who are instructed to evaluate the given images by four properties of textures, i.e., "complexity", "coarseness", "contrast", and "strength". At the same time, their two-dimensional eye-displacements are analyzed by the Maximum Entropy Method(MEM). The MEM is applied to 0.5-second segmented data sets of the 20-second data in order to understand a time-sequential behavior of the eye movements. The results show that eye movements are induced by specific features of the images. It is suggested that this finding should be able to utilize for the objective evaluation in addition to the conventional subjective evaluation.