Converting Night Time International Space Station Images into Lighting Systems Inventories

In this paper, we derive the luminous flux and spectral type of lighting systems of a region from colour images taken on board the International Space Station. On such images, one can only detect the illuminated area on the ground after the emitted photons experienced a reflection process. If one assume a typical reflectance of the underlying ground surface, a Lambertian reflection and a null Upward Light Output Ratio (ULOR) of the lighting device, it is possible to estimate the radiant flux of these lighting devices by integrating its associated illuminated area. Prior to such integration, we evaluated the interest of using a Richardson-Lucy deconvolution algorithm to facilitate the extraction of all light related to each light source. The intensity weighted center of the illuminated area is used to define the final source position. The extraction of the illuminated area is based on a watershed algorithm. Under some assumptions, the integrated photopic radiance may then be converted into light fixture luminous flux. Along with the flux, we derive the spectral type of the lighting device from the color-color ratios. The resulting lighting systems inventory is accurate in terms of positioning sources, but not in terms of retrieving the spectral class of these lighting systems.