Since the beginning of Computer Vision, the resolution of image de-hazing has been a problem. Due to the presence of numerous air particles, resulting in haze, fog, and so on, photographs obtained under unfavorable weather circumstances often seem to be of low quality. This, in turn, makes recognizing objects in a picture difficult. This poses issues for many computer vision problems that depend on picture visibility. The image captured under haze as well as other weather conditions has a process of image deterioration. Image dehazing is a difficult as well as ill-posed task. It overcomes the difficulties of manually constructing haze-related characteristics by using deep learning algorithms. We develop a neural network for image de-hazing in this study. The network model consists of two phases: first, the network is given a foggy image and is tasked with estimating the transmission map; next, the network is given the transmission map estimate and the ratio of the foggy image to the transmission map, and is used to perform haze removal. It avoids estimating ambient light as well as enhances dehazing performance. The haze and dehaze datasets are used as the training set for the proposed scheme. The experimental outcomes for the full-reference metrics SSIM, PSNR, RMSE, MSE, or BRISQUE validate the suggested method's reliability and effectiveness.