An Intelligent EOG System using Fractal Features and Neural Networks

Abstract

The Electrographic signal is measured by moving the eyes from left to right or up and down which create an electrical deflection. EOG signal with the help of eye movement for Parkinson’s patients and voluntary movements are examinedinterms of accuracy, object detection and latency. . In this paper, the EOG data was obtained from five participants and the respective data was preprocessed and box counting feature extraction techniques were used to extract the features. The extracted features were then fed into feed forward neural network. From results, it can be observed that subject 3 performs the functional activity with maximum mean accuracy for all the tasks (down, up, left and right) 90% and also can be observed that subject 1 performs the functional activity with minimum mean accuracy for all the tasks (down, up, left and right) 83.25%.