Research On Image De-fogging Algorithm Based On Variable Frame Retinex

Fog is a very common phenomenon in nature,if the image is in fog and other bad weather to get,then the image will be seriously degraded,the field of life in the image information application caused a great negative impact.Therefore,the study of fog image de-fog technology has important practical significance.In this paper,the main effects of atmospheric aerosol particles on the propagation of light and the physical model of atmospheric degradation are briefly introduced,and then the Retinex algorithm is analyzed emphatically.Retinex is an image enhancement theory based on color constancy.Its core idea is to estimate the illuminance component of the image and then obtain the reflection component of the object,that is,the original appearance of the object,by the original image and the illuminance component.Compared with other image enhancement methods,Retinex algorithm has the characteristics of sharpening and color persistence,so this paper focuses on the analysis.This paper introduces the single-scale Retinex(SSR),multi-scale Retinex(MSR),multi-scale Retinex(MSRCR)and variable frame Retinex with color recovery in detail.According to the experimental results,the advantages and disadvantages of these algorithms in improving image contrast,computational complexity and color fidelity are evaluated.Finally,based on the previous research,this paper improves the Retinex defog algorithm,and the main works and conclusions are as follows:A fast solution model is given.The traditional defog algorithm,in the establishment of the fog model,often process the color image directly.However,after research and analysis,it was found that the fog mainly affected the brightness information in the image,and had little effect on the color tone.Aiming at the traditional fog algorithm,when the model is very complicated,this paper gives a fast solution model.In this paper,the color image is transformed from RGB color space to YIQ space,and then only the Y channel representing the luminance information is defog operated.Finally,the fast Fourier technique and splitting Bregman iteration are used to solve the objective function.The solution model can effectively improve the efficiencyA RLTV-Retinex model,combined with variational regularization terms of RL operator,is given.It is named as two-step TV-Retinex defog algorithm.The effect of RL operator can remove the bright light much better.Combining with the TV-Retinex model,the edge information can be effectively enhanced,and the halo artifacts can be removed.At the same time,the speed of the algorithm is improved dramatically.The experimental results show that the two-step TV-Retinex defog algorithm can better preserve the distortion of the target boundary and the detail information in the image,and remove the halo artifact.In this paper,a defog algorithm based on gradient fitting prior information is optimized,and a method based on gradient fitting prior information TV-Retinex defog algorithm is presented.Based on the gradient fitting a priori information defog algorithm has a strong,scalable features,in the image recovery process can remove the blur in the image,it can be used for image defog areas.Based on the gradient de-ary information defog algorithm,this paper introduces TV-Retinex to optimize the image detail information.The experimental results show that the algorithm can effectively improve the image detail information.