Object Identification from Haze Images using Optimal Adaptive Technique for Light and Dark View
Buvanesh Pandian V, T. Arun Prasath, M. Pallikonda Rajasekaran, G. Vishnuvarthanan · Research Square · 2023
Abstract Images taken in foggy weather frequently have poor visibility. We suggest an effective regularisation technique in this paper to get rid of hazes from an image. An investigation of the transmission function's built-in boundary constraint is very beneficial to our strategy. This restriction is modelled into an optimization problem to estimate the unknown scene transmission. Additionally, a very effective variable splitting-based approach is described to address the issue by transmission refining techniques. The suggested method can restore a high-quality, haze-free image with authentic colours and fine image details with only a few broad assumptions. The usefulness and efficiency of the suggested strategy are demonstrated by experimental results on a range of hazy photographs. Then from the result, five different image quality parameters are identified for both dark and light view to show the quality of the images.