Electrooculography based blink detection to prevent Computer Vision Syndrome

The present work proposes an artificial system capable of preventing Computer Vision Syndrome from the analysis of eye movements. Ocular data is recorded using an Electrooculogram signal acquisition system developed in the laboratory. Wavelet detail coefficients obtained using Haar and Daubechies order 4 mother wavelets are used as signal features. From the recorded data, blinks are classified from any other type of eye movements using Support Vector Machine (SVM) classifier with different kernel functions. We obtain a maximum average accuracy of 95.83% over all classes and participants using second order polynomial kernel SVM classifier. Then the trained classifier has been used in real time to detect blinks. The system is designed to count the number of blinks in a particular interval of time thereby reminding people working on a computer for long periods to rest and blink frequently in case of insufficient number of blinks. We validate the method using a study on ten participants in real time.