Haze Predication Based on Image Quality Score

Abstract

Haze is a prevalent term within the field of image processing, encompassing both naturally occurring phenomena and aerosols generated by human activities. It gives rise to light scattering and absorption, leading to reduced image visibility. This diminished clarity poses challenges for various photographic and computer vision applications, including object recognition and localization. Consequently, there is a growing need for a method to estimate haze density accurately. In this research paper, we introduce a novel model called the "haziness degree evaluator." This model enables the prediction of haze density from a single image, eliminating the necessity for a reference haze-free image. The proposed model quantifies haze density through the optimization of an objective function that encompasses haze-related features derived from correlation and computational analysis.