Driving Realistic Texture in Simulated Long-Wave Infrared Imagery

The visually-realistic yet radiometrically-accurate simulation of long-wave infrared (LWIR) imagery is a problem that has plagued members of industry and academia alike. Simulating texture in LWIR imagery is a more complex task due to temperature variation which is a function of solar absorptivity, surface orientation, shadowing and bulk thermal properties. In order to deal with these additional sources of variability, we have improved the Digital Imaging and Remote Sensing Image Generation (DIRSIG) model to allow the use of thermal property texture maps to control various thermal properties. This paper presents these methods and applies them in order to simulate LWIR imagery taken over natural desert scenes in Trona, CA. The resulting imagery demonstrates that these new methodologies for modeling thermodynamic phenomena and the utilization of an enhanced DIRSIG tool improve the average root mean-squared error (RMSE) of our synthetic LWIR imagery by up to 88%.