We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Real time blink recognition from various head pose using single eye.
- Authors
John, Sofia Jennifer; Sharmila, Sree T.
- Abstract
A vision based human system interface has gained its significance in many areas like drowsy driving, Computer Vision Syndrome (CVS), face detection or recognition etc. The image processing method used in these applications for eye detection is Viola-Jones algorithm. In a natural communication environment, a human face tends to have different roll, yaw and pitch orientation angles. Thus, maintaining a straight focus with a camera is not always practically possible. Motivated by this challenge, experimental analysis is done to identify an eye blink on various head orientation angles. The proposed idea suggests using a single eye to detect the eye state (open or close) rather than both eyes, as blink detection is the rapid closure of both eyes simultaneously and a single eye can be progressively located for wider orientation angles. For analysis roll orientation angles are computed based on eye canthus line and yaw orientations angles are based on nose-line. Accepted experimental results from the customized dataset in different situations shows that the proposed work of using single eye outperforms both eyes with more accuracy and can be resolved for wider orientation angles.
- Subjects
BLINKING (Physiology); HUMAN facial recognition software; ALGORITHMS; PIXEL density measurement; ORIENTATION physiology
- Publication
Multimedia Tools & Applications, 2018, Vol 77, Issue 23, p31331
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-018-6113-3