Measuring Behavior 2005: Letonsaari et al.

Versatile pupil size and eyelid closure video measurement system

M. Letonsaari, J. Virkkala and J. Päällysaho

Department of Clinical Neuroscience, Finnish Institute of Occupational Health, Helsinki, Finland

Usually pupil parameters and eyelid closures are measured by special real time hardware. We have developed a versatile video measurement system for pupil size and eyelid closure registration using low cost consumer electronics. Pupil size variations are used in calculating pupillary unrest index for vigilance analysis. System estimates also blinks and the percentage of eyelid closure which are important parameters for estimating sleepiness.

Video images were recorded with an infrared video camera and with a consumer level digital video recorder. Digital video recorder was used in normal mode and with infrared enabled (‘nightshot’). Both bright pupil and dark pupil methods can be used. Videos were recorded in a laboratory and on field. Video can be recorded on DVD, on tape or direct on computer for computer analysis.

Video is analyzed offline using a computer analysis program. Using offline analysis enables more advanced signal processing making the parameter estimation less vulnerable for errors. Several image recognition algorithms are implemented in the analysis program so that best analysis can be chosen for each video image. Modifications for different types of video can be done with little effort. The analysis is automated but user is also able to inspect automatic parameter estimation frame by frame if needed.

The system is used for vigilance analysis but it can also be used for mental workload estimation. High cognitive workload is known to correlate with dilated pupils. These pupil size changes are very subtle and pupil size estimation accuracy is important. Pupil is therefore estimated with an ellipse and special care is taken for the fitting algorithm accuracy.

The system is ideal for recording pupillography data with minimal hardware resources e.g. in workplaces with offline analysis of the pupil parameters.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.