Dual-band transmittance defogging model

The fusion of shortwave infrared (SWIR) and visible light has emerged in recent years as an effective technique for image defogging. However, in some dense fog scenes, even this fusion method does not achieve significant defogging effects. Some researchers have attempted to apply atmospheric scattering models for defogging fusion images; overlooked is the fact that this model has a key <i>a priori</i> assumption: "the transmittance in the visible light band is assumed to be a constant value." The limitation of this assumption does not apply to multi-band fusion images. Addressing the issue of the model's inapplicability, this paper proposes a dual-band defogging model. First, fusion weights are calculated based on gradient and saturation to generate the fusion image. Subsequently, we calculate the transmittance for each band separately and combine it with the fusion weights to generate the overall weighted transmittance. Compared to the model that generates transmittance from fusion images, the proposed transmittance can better display details in hazy areas and perform well when processing the foreground. The proposed model was compared with three single-band transmittance models and eight shortwave infrared fusion algorithms, proving the effectiveness of the proposed model in subjective visual and objective no-reference image quality assessment (NR-IQA).