Abstract
Purpose :
Numerous studies in humans and animal models have reported that prolonged near work and low light levels are positively correlated with the rate of myopia progression. The average time spent daily with mobile devices has tripled since 2013. Children 8 and under on average spend 2.3h with screen media daily. Modern mobile devices are fully suited to measure illumination and near work distance in natural conditions at high acquisition rates. We test the accuracy of a custom-developed Android app for measuring the illumination of the face, and distance between it and the device for normal, everyday use.
Methods :
Subject’s head was fixed with a chin rest in front of a mobile device (Samsung SM-A520F smartphone with Android 6.0.1 OS) which was mounted in an optical bench where the face-screen distance and face illumination was controlled (Fig. 1). Reference face illumination was measured using the INS DX-100 light meter affixed to the subject’s forehead. Initial calibration was performed to take into account individual difference between face reflectance and a 18% grey reference, and also camera sensor characteristics. The app uses the RGB array of the front camera and custom image processing algorithms tied to physical parameters (ISO, F/#, exposure time, camera field of view and pixel pitch) to measure face illumination and face-screen distance. Face illumination was changed from 20 to 2000lux and distance from 190 to 420mm. Thee measurements were taken each time.
Results :
Linear fit to the measured (calibrated) face illuminance as a function of reference illuminance showed a slope of 0.70 (R2=0.98). The mean error was 20% and mean SD was 5 lux. Measured face illumination depended little on the face reflectance. Linear fit to the measured distance as a function of the real, stepper motor distance showed a slope of 0.98 (R2=0.98). The mean error and SD was 11±10 mm.
Conclusions :
The accuracy of a custom-developed Android app to measure face illumination and face-screen distance was good. The app can work in the background, allowing the subject to use the device as usual, under the broad range daily illumination and distance conditions. The ability to use face detection and tracking to dynamically find the face and set the region of interest will improve measurement accuracy in the future
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.