ABSTRACT
PURPOSE: To describe that Computer image recognition technique is more accurately and less invasive than other methods, which using contact lens with a search probe or electrodes for record of eye movements. METHODS: A reference marker which was infrared was attached to the center of the forehead of the patients and the infrared light was illuminated to both eyes. Video image sequences were recorded using digital CCD camera and the captured image frames (640 X 480) were processed using a image analysis program. From which each image frame captured, pupil area was saperated using difference of brightness in pupil and iris. And then the pupil center was determined by calculating the center-of-mass of black pixels to meet the pupil threshold criteria. RESULTS: In a post-processing process, we could acquire horizontal and vertical eye position and velocity data. Inverse fast fourier transformation(FFT) and digital filtering algorithm were applied to filter out noise due to limited resolution of the CCD camera and sampling rates. Experimental result showed that the system could detect about 0.2mm positional location and 30Hz horizontal and vertical eye movements simultaneously. CONCLUSIONS: We could accurately measure and record two dimensional eye movements of patients (horizontal and vertical simultaneously) in real time with some head movements.