Research On Image Defogging Based On Dark Channel Prior

In recent years, the fog weather appears more and more frequently because of human production and life, which in turn has a negative impact on all aspects of human activities. In foggy weather conditions, the visual visibility is severely degraded due to atmospheric parti-cles having a scattering effect, which is even more so especially in relatively large concentra-tions of fog conditions. In this kind of weather conditions, the collected images are prone to poor contrast and scene information, the loss of details, color distortion and other defects. This greatly limits the application of machine vision in the field of military and national de-fense, remote sensing, satellite imagery, traffic navigation, surveillance systems, target recog-nition. Therefore, the study of image fog removal has a very big significance in practical ap-plications. At present, image fog removal has become a hot field of machine vision research.In this paper, the research of image restoration to defogging is as followings:On the basis of atmospheric scattering model and the propagation characteristics of the fog light, this article deeply studies image to fog algorithm based on dark channel prior and summarizes up the algorithm's disadvantages:first, it is inaccurate for estimation of the at-mospheric brightness value; second, the efficiency of transmission optimizing is very low.As dark channel prior to defog algorithm is inaccurate for estimation of the atmospheric brightness value, this article proposes a new method in this paper for estimation of atmos-pheric brightness value:for the image containing the sky region, the brightness value of the highest frequency in the selected sky area of the dark channel chart is as atmospheric bright-ness value; for the image not containing the sky region, it reasonable sets of gray space in the dark channel chart to determine the approximate range of the atmospheric optical area, in this area which selects the pixel brightness value of the highest frequency as atmospheric bright-ness value.The dark channel priori to defog algorithm uses the method of solving large-scale sparse linear equations to optimize the transmission, which leads to the defogging efficiency is very low. This paper uses the guided filter to optimize the transmission, so that the running time is reduced by about 90% and transmission optimization results are better than dark channel al- gorithm.For assessing the performance of defogging algorithm is inconsistent in the objective and subjective aspects, this paper uses method based on human visual perception for defogging effect evaluation. And color natural index is amended so that the results of quantitative evalu-ation are closer to the conclusion of the human visual perception.