Automated Eye Blink Detection and Tracking Using Template Matching

Eye blink detection is considered to be one of the most reliable sources of communication in modern human computer interaction (HCI) systems. This paper proposes a new method for eye blink detection using template matching and similarity measure. In order to minimize the false detection due to changing background in the video frame, face detection is applied before extraction of the eye template. Golden ratio concept is introduced for robust eye detection and is followed by eye template creation for tracking. Eye tracking is performed by template matching between template image and surrounding region. The normalized correlation coefficient is computed for successful eye tracking. Eye blink detection is performed based upon the correlation score as the score changes significantly whenever a blink occurs. The proposed system provides an overall precision of 92.8% and overall accuracy of 99.6% with 0.1% false positive rate in different experimental conditions.