A Single Image Dehazing And Denoising Algorithm That Combines The Dark Channel Prior And The Variational Model

Image dehazing is one of the important contents of digital image processing.Traditional dehazing methods can only remove the haze but ignore the noise in the image.The noise what is covered by the haze is highlighted while removing the haze.This situation seriously affects the visual effect of the image.Focusing on the above problem,a variational method for single image dehazing and denoising combining dark channel prior is proposed.Firstly,we estimate the transmission map using dark channel prior(DCP),then design a series of new variational models for color image dehazing and denoising based on this estimation and the regularizer of variational models.In order to improve the computation efficiency of the series of models,we design their fast split Bregman algorithms.The main works include following aspects:(1)We propose three new local variational models based on the transmission map and the layered total variation(LTV)regulariser,multichannel total variation(MTV)regulariser,and color total variation(CTV)regulariser,respectively.Numerous experiments are presented to compare their denoising effects,edge-preserving properties,and computation efficiencies.Numerical results further prove that above models not only can effectively remove the haze and noise in the haze image simultaneously,but also has a good edge preserving effect.(2)We propose a new non-local variational models based on the transmission map and the non local color total variation(NL-CTV)regulariser.And we compare the NL-CTV model with the above three models.Numerical results prove that the NL-CTV model is better than the above three models in image dehazing,denoising and edge preserving.(3)To demonstrate the merits of the proposed method,we also conduct some comparisons with several existing state-of-the-art methods.It is found that the image restored by our method has excellent dehazing effect and denoising effect,and effectively enhances the contrast of the image after the objective evaluation.