Comparison and analysis of deep learning for Dehazing

Abstract

Hazy image is caused by environmental factors such as uneven air light, darkness, contrast, saturation, and attenuation. Due to limited visibility, various applications such as ITS, tracking systems, object detection devices, and others may fail. Image dehazing is used to minimise the effects of weather conditions and improve the pictorial impacts of the image which make post-processing easier. Various approaches are utilized to create high-quality images with improved colour, lighting, and boundaries which are free of halo artifacts. To eliminate the haze, most known de-hazing algorithms employ a conventional scattering model, prior’s methods. This research compares a current de-hazing method based on deep learning to existing state-of-the-art technologies.