Camera-based visibility estimation: Incorporating multiple regions and unlabeled observations

a r t i c l e i n f o This paper investigates image processing and pattern recognition techniques to estimate atmospheric visibility based on the visual content of images from off-the-shelf cameras. We propose a prediction model that first relates image contrast measured through standard image processing techniques to atmospheric transmission. This is then related to the most common measure of atmospheric visibility, the coefficient of light extinction. The regression model is learned using a training set of images and corresponding light extinction values as measured using a transmissometer. The major contributions of this paper are twofold. First, we propose two predictive models that incorporate multiple scene regions into the estimation: regression trees and multivariate linear regression. Incorporating multiple regions is important since regions at different distances are effective for estimating light extinction under different visibility regimes. The second major contribution is a semi-supervised learning framework, which incorporates unlabeled training samples to improve the learned models. Leveraging unlabeled data for learning is important since in many applications, it is easier to obtain observations than to label them. We evaluate our models using a dataset of images and ground truth light extinction values from a visibility camera system in Phoenix, Arizona. Atmospheric visibility can be a useful indicator of atmospheric pollution resulting from suspended particulates especially in drier climates. This coupled with the rapidly growing number of cameras in our ecosystem motivates image-based visibility estimation as an appealing complement to traditional means of monitoring air pollution since specialized equipment for measuring pollution is comparatively expensive. So-called visibility camera systems are already seeing widespread deployment. For example, the Interagency Monitoring of Protected Visual Environments (IMPROVE) 1 program has installed and maintains cameras in over two dozen national parks in the United States. In addition, regional air quality agencies 2 have deployed visibility camera systems in over 30 cities. More broadly, though, there are potentially tens of thousands of web, surveillance, traffic, and other cameras, which could be used to monitor atmospheric visibility and thus air pollution. The work in this paper represents a step towards using multimedia data, in particular images from off-the-shelf cameras, to perform quantitative estimation of atmospheric visibility. We investigate image processing and pattern recognition techniques to derive prediction models of light extinction based on image content. Light extinction captures the joint effects of light scattering and absorption that result from particulates in the atmosphere. Our major …