A Physics-based Fusion Approach for Color Night Vision

Night vision enhancement systems, such as night vision goggles and in-vehicle displays, have been used extensively in the military since the 1960’s. Most common night vision systems use single-band sensors that are sensitive to either emitted infrared (IR) radiation or reflected light, and thus provide a monochromatic image (intensities of grey or green) of the inspected scene. As such, depth perception is limited and visual acuity and focal range are restricted. Adding color can increase depth, contrast and perception of night vision compared with monochromatic images, improving soldiers' reaction time and reducing fatigue. We propose a novel physics-based approach to fusing images in multiple IR wavelength bands that should provide reliable and consistent color assignment and generate color images close to natural appearance for color night vision. The approach, which features image restoration, material identification, and color mapping, will utilize a visual color knowledge base generated from experimental multispectral and hyperspectral data and first-principles radiation transport simulations. The results from Phase I will be used to define real-time algorithms for color night vision applications in Phase II.