Joint modeling with eye movement and pupil scaling for affective assessment based on kpca algorithm

摘要

Aiming at the fact that psychological state can be reflected by eye movement and pupil size, an affective assessment method based on the jointmodel of eye movement trajectory and pupil scaling was proposed in this paper. Firstly, the experimental apparatus was developed to capture and transmit the eye images. Secondly, multiple advanced image processing algorithms were synthetically adopted to extract the pupil. Both the position and size of the pupil were obtained. Thirdly, the joint model with the eye movement trajectory and pupil scalingwas constructed as feature vector, which was subsequently processed with kernel principal component analysis (KPCA) algorithm to reduce its dimension. Finally, the nearest neighbor classifier was built according to the dimensionality reduction information to implement the classification of the samples.With proper experimentalmethod designed for collecting samples, this approach can be used for affective assessment. Experimental results had demonstrated the good practicability of our study.