Enhancing Visibility: A Novel Approach for Dehazing Aerial Images with MixDehazeNet
S.C. Saxena, S Kra ̈mer, Arun Singh Pundir · 2024
The demand for high quality aerial images is pressing because of their diverse applications in aircraft navigation, geological analysis, military reconnaissance, cartography, and other fields. Haze consists of suspended particles in the atmosphere that can lead to a deterioration in image quality, hence affecting the precision of more advanced applications. There is a plethora of material accessible on the topic of removing haze in ground imagery, but there is a scarcity of methods expressly designed for aerial photos. Furthermore, existing dehazing techniques lack the ability to perform real-time processing. This paper presents a technique for removing haze from aerial photos using MixDehazeNet. The model underwent training using a well selected dataset consisting of a wide range of aerial photos taken from different heights and geographical areas. This approach guarantees the model’s resilience in many environments and circumstances. The modified network yielded enhanced dehazing outcomes. The results of our experiment demonstrate that the modified MixDehazeNet model applied to our aerial image test yields a Peak Signal-to-Noise Ratio (PSNR) of 36.89 decibels (dB).