Measuring Atmospheric Scattering from Digital Images of Urban Scenery using Temporal Polarization-Based Vision (1407.3540v1)

Abstract: Particulate Matter (PM) is a form of air pollution that visually degrades urban scenery and is hazardous to human health and the environment. Current monitoring devices are limited in measuring average PM over large areas. Quantifying the visual effects of haze in digital images of urban scenery and correlating these effects to PM levels is a vital step in more practically monitoring our environment. Current image haze extraction algorithms remove haze from the scene for the sole purpose of enhancing vision. We present two algorithms which bridge the gap between image haze extraction and environmental monitoring. We provide a means of measuring atmospheric scattering from images of urban scenery by incorporating temporal knowledge. In doing so, we also present a method of recovering an accurate depthmap of the scene and recovering the scene without the visual effects of haze. We compare our algorithm to three known haze removal methods. The algorithms are composed of an optimization over a model of haze formation in images and an optimization using a constraint of constant depth over a sequence of images taken over time. These algorithms not only measure atmospheric scattering, but also recover a more accurate depthmap and dehazed image. The measurements of atmospheric scattering this research produces, can be directly correlated to PM levels and therefore pave the way to monitoring the health of the environment by visual means. Accurate atmospheric sensing from digital images is a challenging and under-researched problem. This work provides an important step towards a more practical and accurate visual means of measuring PM from digital images.