Abstarct: With the advent of technology and extensive advances in rehabilitation techniques, research is being developed to use reliable, low-cost, and easy-to-use devices. Among the techniques of rehabilitation, research on human-machine-to-machine interactions (HMI) can be a great help for physically disabled people. This study examines and analyzes eye movements as a new method for identifying the activities of individuals. Eye movement information is obtained using the electrocautogram signal recording system (EOG). One of the easiest signals that can be used to help people with disabilities is the EOG signal. Also, using this signal, you can detect the behavioral pattern and the way people deal with different issues. The study examines eye movements using signals that are recorded in three classes, including rest, reading articles and web browsing. In this study, the pre-processing is performed first to reduce the noise and determine the appropriate signal range. Then, using the threshold values, the type of motion of the eye is determined. Sections of saccades, Fixation and blxink are segregated. Then, by extracting the temporal characteristics of these areas, the eye movement is evaluated when performing activities. In this study, we extracted two new features called Hermit Transform coefficients and linear prediction coefficients from Saccades and Fixation areas, by adding these features we found that the first feature improves performance and the latter does not affect performance improvement. The accuracy of the detection for 80 features is 64% and for the 28 features is 71%. These precisions have been increased to 79% by adding the coefficients of the Hermit Transform for 28 features and 67% for 80 features.