The Research Of Image Fusion Haze Removal Algorithm Based On Adaptive Correction

Single image defogging has become an important research in many high-level computer vision tasks,which is a challenging problem.Because the transmission map is difficult to estimate,and the single image defogging algorithm may be affected by noise amplification in the sky area,the defogging algorithm based on multi-scale fusion can effectively solve this problem.However,this method cannot effectively deal with the problem of image color distortion.Based on this,this paper analyzes the cause of haze image degradation and blurring mechanism,improves the existing image defogging algorithm and introduces new ideas.First,according to the difference in optical components of fog and haze,an image dehazing algorithm based on Adaptive Correction(AC)is proposed.Then,according to the characteristics of the image of the coal burning flame in the rotary kiln affected by the light source,an image dehazing algorithm based on multiple exposure fusion is proposed.Many experiments were conducted at the same time,and the specific research results are as follows:(1)This paper designs a multi-scale visual enhancement module.First,use gamma correction,local contrast enhancement,and average contrast enhancement to generate three input images.Then,three weight maps of contrast,dark channel,and saturation are extracted as the weight maps of the input image.Finally,the multiplicity based on the Laplacian pyramid is used.The Multi-Scale Fusion(MSF)strategy fuses the input image and solves the problems of loss of detail,reduced contrast and saturation,and abnormal brightness in haze areas.At the same time,in view of the unsatisfactory effect of the existing defogging algorithms in processing color-distorted haze images,this paper also proposes a multi-scale color balance module.First,we use white balance and Contrast-Limited Adaptive Histogram Equalization(CLAHE)to generate a color-corrected input image,then export two weight maps of brightness and semi-inversion,and finally use the MSF strategy to fuse the input image to solve the haze image color distortion problem.(2)The existing defogging algorithm cannot adapt to changes in the environment and properly calibrate the parameters of the algorithm.Therefore,improving the robustness of the algorithm is a bottleneck of the defogging algorithm.In order to improve the algorithm's resistance to the influence of different environmental factors while ensuring the complexity of the algorithm,this paper proposes an adaptive correction(AC)algorithm.First introduce the white balance image,define the color deviation image,then use the maximum value of the R and G color channels in the deviation image to describe the degree of color distortion of the haze image,and finally the dehaze image after multi-scale visual enhancement and multi-scale color balance are combined adaptively.The adaptive combination can compensate for the color distortion caused by the haze image,and effectively improve the visual effect of the haze image.(3)the kiln flame images in industrial rotary kiln are blurred by light source and dust,which seriously affects the follow-up work of the kiln flame images.In view of the shortcomings of the existing defogging algorithm in image processing of the kiln flame images,this paper proposes an image defogging algorithm based on multi-exposure fusion.First,we use gamma correction to adjust the brightness of the image to obtain a series of artificially exposed images,and then use the MSF strategy to fuse the obtained multiple exposure images into a fog-free image.Experimental results show that the kiln flame images restored by the algorithm is clearer and more natural,and our algorithm is conducive to the identification of each area of the flame image.