City-scale distance estimation via near-infrared trispectral light extinction in bad weather

This paper proposes a novel physics-based model for city-scale relative distance estimation using atmospheric optics. The suspended particles in the atmosphere can attenuate nearly every wavelength of the visible spectrum, particularly when the particle concentration is high. In this regard a trispectral distance estimation model is derived by utilizing the extinction coefficient differences between three selected near-infrared wavelengths. It may be noted that the proposed approach eliminates the influence of the surface materials (of the buildings) by linearly interpolating the reflectance spectra. As the atmospheric extinction coefficient is related to the wavelength and meteorological conditions, the visibility information is also utilized to make the proposed model simple and effective. A trispectral imaging system, constituting of a monochrome sensor, three narrow-band filters and an automatic filter rotation device, is utilized to capture the data across three near-infrared spectral bands. Experiments conducted in this study validate the superior performance of the proposed model for city-scale distance estimation.