Image Defog Algorithm Based On Prior Knowledge Of Dark

Time is progressing,technology continues to develop,at the same time,the phenomenon of haze pollution is more and more serious,to people's daily life and work,etc.have brought a great impact.At present,the existing dark primary color de-fog method in the image fog has achieved good results,however,using this method to deal with the image,there are some white halo at the edge of the block,and the entire algorithm of time and space Complexity is also very high.In this paper,we study the degradation mechanism of the fog image on the basis of the dark primary a priori,and propose two improvement methods for the shortcomings of some special areas such as white scenery and sky area.First,reduce the pixel value of each dark primary channel and then optimize the transmittance;Second,the dual-region filter and guided filtering combined to repair the transmittance.It is found that the two improved methods not only significantly reduce the running time,but also eliminate the white halo block effectively compared with the soft dug method in the original dark primary method.So that the image after the fog as a whole look more clear and natural.The specific research contents are summarized as follows:1.Determine the sky area in the fog image.The dark primary a priori theory does not apply to the white area in the image,so before the fog image is judged,the sky area included in the image is first separated and the sky image of the target image is obtained from the sky image of the target image The required atmospheric light value.According to the location of the sky area,color and other characteristics,to obtain its binarized image and then morphological processing,and finally accurately split the image of the sky area.2.An improved method to reduce the pixel values of three dark channels is proposed.Dark primary color in the first three color channel pixel values are relatively high,will affect the fog effect,so consider the three color channel values were reduced to make it tend to zero,improved after obtaining three dark channel pixel values are Very small images,de-fog them to restore,and finally weighted to obtain the original clear image.It is found that the method has better effect on the fog image with white region than that of the existing method,and can effectively reduce the noise of the existing method in this case and improve the visual effect.3.An improved method of combining dual-region filtering and guided filtering is proposed.In order to eliminate the white halo block at the same time,we can reduce the time consumed by this method.The method of combining the double region filter to repair the transmittance and the method of guiding the filter to repair the transmittance are used to solve the target image transmittance.In combination with the advantages of both at the same time,but also make up for their own deficiencies.Compared with the original method,the use of soft matting to repair the transmittance method significantly improves the efficiency of image processing and has a better processing effect.