The presence of haze in the atmosphere often results in low contrast and limited visibility in images and videos taken in hazy weather. Environments that are not favourable to photographs, such as fog, ice fog, steam fog, smoke, volcanic ash, dust, and snow further degrade and degrade the color and contrast of photographs. Due to degraded images, many computer vision applications fail to work effectively. Various computer vision applications, such as photometric analysis, object recognition, and target tracking, are unavoidably degraded by hazy images and videos with biased colour contrasts and poor visibility. By using image haze removal techniques to remove those deficiencies, computer applications can be used for both image comprehension and classification. In the field of image dehazing, deep learning methods are now being applied. There have been many studies done on removing haze. Reviewing works on dehazing daytime and night-time images, this paper discusses several approaches for removing haze and describes them all in detail. A statistical method or a network-based method refines an image by using statistical observations of the scene. In order to compare their dehazing potential, these algorithms were assessed quantitatively and visually. This paper presents a comparison of the various methods for image and video dehazing that use different methods for estimating transmission and atmospheric light. Signal-to-noise ratio (SNR) and structural index similarity (SSIM) are two of the tools used to assess image quality. In our study we will consider database such as RESIDE, SceneNet, I-HAZE, O-HAZE, D-HAZY, Middlebury, NYU Depth Dataset V2.