Characterization of a quasi-real-time lighting computing system based on HDR imaging

In this paper we present the characterization of a calibrated embedded system for quasi real-time lighting computation based on HDR sky monitoring. To quantify its accuracy in lighting computation, we positioned the device in front of a test module with unilateral-façades to measure the luminance distribution of the sky and ground dome, including that of the sun, the clouds and landscape, followed by on-board illuminance computation in the device. An experiment was conducted on two representative days, with overcast and clear sky respectively. The embedded device computed the illuminance for 5 virtual sensors which was referred to the measurement by 5 lux-meters at the identical positions synchronously to assess its relative error. Compared with the Perez all-weather sky model, the embedded device was indicated to be more reliable and achieved 10%-25% higher accuracy in transient lighting computation of horizontal illuminance based on HDR imaging and luminance mapping.