Oculomotor Plant Feature Extraction from Human Saccadic Eye Movements

Eye movements can be used as a source to predict the user interest behavior in human computer interaction system. Oculomotor Plant Feature (OPF) values are the anatomical components of extra ocular muscles which are responsible for the eye movements. This paper is focusing on extracting the optimized oculomotor plant feature values from the saccadic trajectories generated by the two-Dimensional Oculomotor Plant Mathematical Model (2DOPMM) and selecting the best optimization algorithm to extract the optimized OPF values. The root mean squared error is used to minimize the error between simulated and classified saccadic trajectories and the existing algorithms such as Nelder-mead, levenberg-marquardt and trust-region reflective algorithms are compared to analyze the processing time for the optimization. This project concludes by analyzing results of optimized OPF values and the optimization algorithms used.