Application Of Random Walk Model In Eye Tracking Protocols

Trajectory data contains abundant information.In our reading,tourism,transportation,social networks and other activities,we are likely to see a variety of trajectory data that contains rich information.With the rapid development of Internet technology,storage technology,virtual reality and other technologies,people's daily activities have produced massive and various trajectory data.It has promoted the prosperous development of the excavation technology.Random walk based model is a mathematical and statistical model used to characterize the trajectory of irregular motion.Based on the random walk model,we can calculate the topological correlation between nodes and solve the problem of node ranking.We can translate many of the problems with the trajectory data into the issues on the graph model.Therefore,we use the random walk model as a tool to study its application in trajectory analyses.Based on the eye movement,we proposed a new method utilizing random walks to identify the fixation center.The extraction of fixation center plays an important role in optimizing human-computer interaction experience In this paper,we proposed a novel framework based on random walks with an adaptive damping factor,to solve the problem of fixation center extraction in eye movement.The method not only depends on the connectivity of the fixation points,but also considers their density characteristics,and weakens the noise of the fixation point.The method firstly extracts and clusters the raw eye movements.Each cluster represents a region of interest of the user.Then,for a particular cluster,each fixation point is marked with an initial score based on its density,and iteratively updated using the random walk model.During each walk,the propagation rate of the fixation point is adjusted adaptively.Thereby the noise contained in the low scoring fixations declined.The main advantages of our proposed method can be summarized as follows:(1)We assigned the fixation point with an initial score utilizing its density.(2)The random walk model can utilize the connectivity among the fixation points,which denotes the global distribution.(3)By iteratively updating the score of the fixation point,the adaptive damping factor will reduce the effect of the fixation with low score.Therefore,our proposed method can obtain a stable and accurate fixation center.A large number of experiments showed that the proposed algorithm is superior to the existing algorithms for fixation center extraction.