Eye Blink Detection Using Variance of Motion Vectors

Abstract

A new eye blink detection algorithm is proposed. It is based on analyzing the variance of the vertical motions in the eye region. The face and eyes are detected with a Viola-Jones type algorithm. Next, a flock of KLT trackers is placed over the eye region. For each eye, region is divided into 3 × 3 cells. For each cell an average "cell" motion is calculated. Simple state machines analyse the variances for each eye. The proposed method has lower false positive rate compared to other methods based on tracking. We introduce a new challenging dataset Eyeblink8. Our method achieves the best reported mean accuracy 99 % on the Talking dataset and state-of-the-art results on the ZJU dataset.

Eye blink detection based on motion vectors analysis

“…The authors do A C C E P T E D M A N U S C R I P T not mention how the false alarm is calculated and in particular how the Negative count has been defined. As Drutarovský & Fogelton [16] mention, the results can differ significantly, if the number of frames with opened eyes is used instead of the number of non-blinks.…”

“…Based on the review of the mentioned methods, we focus on the motion based method to achieve high invariance to appearance clutter such as glasses, head pose or facial expression. We build our method on motion analysis [16] to avoid head movement compensation.…”

“…Lucas-Kanade tracker is also used by Drutarovský & Fogelton [16] to track the eye region. 6 motion vectors are estimated from the eye region and an average motion and variance are calculated to be the input waveforms for a state machine.…”

“…Other methods compute a difference between frames (pixels values [14], descriptors [15], etc.). We built our algorithm upon the successful method of Drutarovský & Fogelton [16] which is a motion tracking state-of-the-art method with very low false positive rate.…”

Liveness Detection from Real user, Printed Pictures and Pictures on Mobile Devices from Low Resolution Webcam

“…Motion-based method; several research studies have employed the optical flow technique to track eyelid motion (Drutarovsky and Fogelton, 2014). In addition, other approaches employ the variance of intensity between the sequenced frames based on threshold to detect blinking (Divjak and Bischof, 2009).…”