Research On Image Enhancement Algorithm Of Haze Weather Based On Retinex Theory

In recent years,with the change of global climate,the frequency of haze weather is getting higher and higher.Because the image acquisition equipment is easily affected by the environment,the image collected in this environment will appear contrast reduction,detail blur,color distortion and other problems.Image,as the main medium of information transmission,has been widely used in the fields of transportation,inspection,aerospace and so on.The acquisition of images in haze weather has brought great influence to the work in these fields.Therefore,it is very necessary to propose an effective image defogging algorithm.In this paper,based on the in-depth study of image enhancement algorithm,combined with the characteristics of haze image itself,Retinex algorithm was studied.The Retinex algorithm mimics the development of the human visual system.Retinex algorithm is widely used in image enhancement because it can perform adaptive enhancement on different types of images.However,there are some phenomena such as halo artifact,insufficient edge sharpening and over-exposure of highlight area in the processed image.On the basis of understanding the status quo of enhancement algorithms and mastering the classical enhancement algorithms,this paper proposes a haze image enhancement algorithm based on Retinex theory,aiming at the existing problems in haze image enhancement.Firstly,the algorithm selects the YIQ color space,and decomposes the low frequency and high frequency components by wavelet transform of the Y component in the color space.Then the improved defogging algorithm is selected to effectively remove the fog in the low-frequency component and improve the clarity of the image.Then a new threshold function is selected to reduce the noise in the high frequency component and improve the image quality.Finally,the image is modified by increasing the global contrast and compensating the saturation component to improve the image enhancement effect.This paper contains the following two innovations:(1)Aiming at the problems of halo artifacts and blurred details in haze images,a Retinex de-fog algorithm based on wavelet domain is proposed.First of all,the algorithm selects the appropriate color space,and only decomposes the luminance component in the color space by wavelet.Then the bilateral filter is improved by using the radiation difference in the pixel range domain,and it is applied to the Retinex algorithm.Finally,the improved Retinex algorithm is used to defog the low frequency components.The algorithm is targeted to remove fog and effectively improves the clarity of the image.(2)Aiming at the noise problem in haze image,a denoising algorithm based on wavelet domain threshold function is proposed.By analyzing the problems existing in the traditional soft and hard threshold functions,the algorithm proposed a threshold algorithm which can adjust the parameters flexibly to realize the random switch between the soft and hard threshold functions,and overcome the shortcomings of the soft and hard threshold functions.Finally,global contrast enhancement and saturation component compensation are applied to the image to make the processed image more conducive to visual observation by human eyes.In summary,this paper proposes a haze image enhancement algorithm based on Retinex theory,and experiments are carried out on the actual images with fog and noise.The experimental results are analyzed qualitatively and quantitatively from both subjective and objective aspects,which shows the effectiveness of the haze image enhancement algorithm proposed in this paper.