A Review on GAN based Image Dehazing
Kavinder Singh, Vishruth Khare, Vishwas Agarwal, Sourabh Sourabh · 2022 6th International Conference on Intelligent Computing and Control Systems (ICICCS) · 2022
The improvement of deep learning facilitated the evolution of computer vision methods. Image dehazing is one of the challenging tasks in computer vision. Haze is a very common phenomenon arising due to bad weather, poor lighting, pollution, etc. Haze poses the problem of poor visibility in images under extreme circumstances. Several deep learning-based methods have been used to overcome haze-oriented problems. Generative networks (GANs) focus on the distribution of the normal images to remove the haze from a hazy image. The methods enhance the performance of computer vision under hazy conditions. This paper focuses on discussing a few important techniques, specially GANs, to solve the problem of dehazing. Various techniques have been qualitatively analyzed and the merits and demerits of the respective models have been discussed. Finally, some light is thrown upon the real-world applications and significance of dehazing.