Analysis of color rendering transformation effects on dehazing performance

Dehazing is an image restoration technique generally applied as a post-processing step in a camera processing pipeline. The performance of dehazing algorithms is strongly dependent on the electro-optics transfer curve of image acquisition. In order to reduce color distortion on restored images, we investigate how color rendering transformations affect the dehazing algorithms in a camera processing pipeline. First, we develop numerical analysis on how a hazy image model is modified by color rendering transformations in a pipeline. Based on the analysis and observations, we conclude that better dehazing performance can be achieved when a dehazing algorithm is applied before color rendering transformations. In addition to the theoretical analysis, experimental results validate our conclusion.