Deep Light Pollution Removal in Night Cityscape Photographs

arXiv cs.CV / 4/13/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper targets the specific problem of urban light pollution in night cityscape photos, where artificial lighting causes skyglow, halos, and washed-out stars.
It introduces a physically-based degradation model that extends prior nighttime dehazing approaches by accounting for anisotropic spread from directional sources and skyglow from hidden surface lights behind skylines.
To address limited paired real-world data, the authors propose a training strategy that combines a large generative model with synthetic-real coupling to improve generalization.
Experiments report substantially reduced light-pollution artifacts and better recovery of authentic night imagery compared with earlier nighttime restoration methods.
The work positions light pollution removal as distinct from classic nighttime dehazing, emphasizing restoration of the “pristine” radiative footprint rather than only air-related visibility enhancement.

Abstract

Nighttime photography is severely degraded by light pollution induced by pervasive artificial lighting in urban environments. After long-range scattering and spatial diffusion, unwanted artificial light overwhelms natural night luminance, generates skyglow that washes out the view of stars and celestial objects and produces halos and glow artifacts around light sources. Unlike nighttime dehazing, which aims to improve detail legibility through thick air, the objective of light pollution removal is to restore the pristine night appearance by neutralizing the radiative footprint of ground lighting. In this paper we introduce a physically-based degradation model that adds to the previous ones for nighttime dehazing two critical aspects; (i) anisotropic spread of directional light sources, and (ii) skyglow caused by invisible surface lights behind skylines. In addition, we construct a training strategy that leverages large generative model and synthetic-real coupling to compensate for the scarcity of paired real data and enhance generalization. Extensive experiments demonstrate that the proposed formulation and learning framework substantially reduce light pollution artifacts and better recover authentic night imagery than prior nighttime restoration methods.