Image Haze Removal Using Dark Channel Prior And Bright Channel Prior

Nowadays,it is often get some hazy images when taking photos in outdoor,and the hazyimages not only reduce the quality of images,but also make the color of the images distorted.Whether in real life or in the filed of computer vision,the influence of haze on images has brought great trouble to people.In recent years,the problem of images dehazing has become more popular in the field of computer vision.In this paper,a new method of images dehazing is proposed based on the dark channel prior.In addition to using multiple images to remove haze,many methods of removing haze from single image have also been proposed.Tan et al.restore clear images by maximizing the local contrast of the input images.The output images are visually convincing,but it is no physical meaning.Fattal et al.assume the transmission and the surface shadow of the object are independent in the local area.The transmission of the media is estimated by estimating the albedo of scene,and the output image is finally obtained.Although the output result is very satisfactory,the algorithm is not applicable when the assumption is not valid.He et al.used dark channel prior to estimate the transmission,and achieved good results.It is a classical haze remove algorithm.However,it still has limitations with high brightness images,the results of dehazing are not satisfactory.This paper is improved on the basis of dark channel prior.First,we find that the bright channel prior is also applicable to the hazy image;then,on the basis of the dark channel prior,the bright channel prior is added and the bright channel prior is combined with the dark channel prior to form a new prior.Finally,a new prior is used to get more accurate transmission,so as to restore haze free images.Experiments show that the proposed method achieves good haze removal effect both synthetic hazy images and natural hazy images.