Accuracy of Six Interpolation Methods Applied on Pupil Diameter Data

Abstract

Eye-tracking is a method of recording the location of gaze as well as pupil diameter (dilation) during active visual behavior. Due to blinking and noise in the recording system, these signals are often briefly “lost”, leading to missing data. Here, we aim to analyze the accuracy of six interpolation methods to complete missing values from pupil diameter data. Data interpolation is a type of estimation that constructs new data from existing, neighboring values. Having the possibility to choose from different types of methods, the question is which interpolation method is the most suitable for pupil diameter data. Here, we applied the linear method, the previous neighbors method and four cubic interpolations. Using real data recorded during a visual eye-tracking experiment, we compared the maximum deviations of each interpolation method. Results indicate that for pupil diameter data, the most accurate reconstruction of pupil loss is obtained by Akima, Makima, and the Piecewise Cubic Hermite Interpolating Polynomial.