Thermal HDR: Applying High Dynamic Range Rendering for Fusion of Thermal Augmentations with Visible Light

Abstract

In safety applications for fields such as navigation or industrial manufacturing, thermal augmentations have often been used to help users safely navigate low-light environments and detect obstacles. However, one problem with thermal imaging is that it often occludes the environment and coloration that would otherwise be useful, for example traffic sign text or warning labels on industrial equipment. In this paper, we explore a new algorithm that takes advantage of high dynamic range (HDR) rendering techniques in order to more effectively present thermal information. Unlike previous fusion algorithms or conventional blending techniques, our setup makes use of both a thermal camera and HDR frames to synthesize a final overlay. Moreover, we set up a series of two experiments with a simulated heads up display (HUD) to 1) measure reaction times to the sudden appearance of pedestrians, and 2) conduct circuit repair. Results showed that the HDR algorithm was subjectively preferred to other approaches, and that performance on average could match other conventional algorithms on most occasions.