August 2012
Volume 12, Issue 9
Vision Sciences Society Annual Meeting Abstract  |   August 2012
PERCLOS Threshold for Drowsiness Detection during Real Driving
Author Affiliations
  • Sheng Tong Lin
    DSO National Laboratories, DMERI, CPP2, Singapore
  • Ying Ying Tan
    DSO National Laboratories, DMERI, CPP2, Singapore
  • Pei Ying Chua
    DSO National Laboratories, DMERI, CPP2, Singapore
  • Lian Kheng Tey
    DSO National Laboratories, DMERI, CPP2, Singapore
  • Chie Hui Ang
    ST Kinetics, WVDTC, Singapore
Journal of Vision August 2012, Vol.12, 546. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sheng Tong Lin, Ying Ying Tan, Pei Ying Chua, Lian Kheng Tey, Chie Hui Ang; PERCLOS Threshold for Drowsiness Detection during Real Driving. Journal of Vision 2012;12(9):546.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

PERCLOS (proportion/percentage of time in a minute that the eye is 80% closed) has been a popular topic of research for assessing driver's alertness. However, most of its validation to date was done in a simulated driving environment. In this study, we exposed drivers to monotonous 30km/hr driving in a closed road circuit for up to 4 hours. Smart Eye Anti-Sleep eye tracker system (SMART EYE AB) was used to capture drivers' eye closure throughout driving. Preliminary analysis revealed that the drivers can be classified into 3 groups. Group E (n = 11) refers to the ‘elites’ who lasted the full 4 hours. In Group V (n = 13) are the drivers who are vulnerable to fatigue driving and drove out of lane within 4 hours. Group U drivers (n = 14) are unmotivated drivers. For each driver, 3 PERCLOS readings were extracted: at the start of the driving as the baseline, at the point of 1-second microsleep and at end point which is upon driving out of lane or completion of the 4 hour driving. There is a significant difference between the 3 PERCLOS readings (F(2,70)=10.79, P <0.01). There is no interaction effect from grouping (F(4,70) = 1.93, P = 0.12). Pairwise comparison ( Bonferroni corrected) between estimated marginal means(EMM) showed a significant PERCLOS increase from baseline (6.13%) for both 1-second microsleep and end point (P <0.01) which is 10.54% and 12.22% respectively. However, there is no significant PERCLOS difference between microsleep point and end point (P = 0.13). This study suggest PERCLOS threshold can be set at 10%, assuming critical point to be 1-second microsleep.

Meeting abstract presented at VSS 2012


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.