Recent advances in image dehazing: Formal analysis to automated approaches

Images captured in hazy environments need to be processed to increase their contrast and colour integrity. Dehazing, sometimes referred to as haze removal is an important pre-processing step for image processing and computer vision applications. Numerous methods for dehazing images have been proposed in the literature. This study provides a complete evaluation of numerous image dehazing techniques and their notable standards. This study extracted and presented an important direction of the many algorithms to handle the challenges of dehazing such as model-based methods, transform domain methods, variational-based algorithms, learning-based algorithm and transformer-based algorithms. This study attempts to compile and evaluate the most important studies in the domain of image dehazing. A variety of factors have been considered necessary to provide a detailed information in this study. These factors include datasets that utilised in the literature, challenges faced by the prior researchers, motivations, and recommendations for reducing the drawbacks in the available literature. The systematic rules are utilized for searching all relevant papers on image dehazing using several keyword diversities along with a glance for assessment and the benchmark studies. Image dehazing, which generally eliminates undesirable pictographic effects is often considered as an image enhancement method. A completely automated process, a valid assessment strategy, and databases based on diverse settings are needed for it to operate under real-time applications. Many relevant studies are conducted in order to achieve these substantial goals. We examined numerous image dehazing methods and assessed the objectivity of the results. The results of our study precisely reflect numerous observations on image dehazing regions in contrast to other review articles. We believe that the findings of the study can be a helpful set of recommendations for professionals looking for a full understanding of image dehazing. In addition, we present a thorough examination of the methods used to evaluate image quality using the full-reference category and the no-reference category. Several standard evaluation metrics are utilised to compare the results of the well-known dehazing techniques. A list of standard haze image datasets is also included in this study so that different dehazing methods can be compared. In our opinion, the findings of this study could act as a valuable guide for experts in quest of a detailed understanding of image dehazing.