Variational Model Using Higher Order Differential Operators And Its Application In Scene Recovery

The images we obtain in daily life are often in harsh environments such as dust,under-water,and haze,resulting in images that are not a complete reflection of the actual scenery.This is because during the process of image generation,transmission,and storage,the image quality will decrease.This process of reducing image quality is called image degradation,which involves changing the true color of the image or hiding some details of the image.Scene restoration is a processing technique that improves image quality by reconstructing degraded image content to maximize its restoration to the original state of the scene.The commonly used scene restoration methods include image enhancement,image restoration,image encod-ing,image compression,etc.In recent years,scene restoration has received increasing attention and has achieved significant results in the field of single image restoration.However,a large number of studies based on dark channel prior(DCP)only estimate the transmission mapping and are only applicable to single scenes for degraded image restoration.Different from pre-vious methods,this paper focuses on the scattering mapping of global ambient light,and the proposed method is suitable for multiple scene restoration.Unlike traditional single scene restoration methods,Rank One Prior(ROP)focuses on the scattering mapping of ambient light,estimating the illumination conditions for degraded imaging,and characterizing scene restoration as a process of achieving degraded imaging illu-mination condition restoration.Therefore,scene restoration methods based on rank one prior can be applied to various scene restoration.Based on the rank one priori scene restoration method,this paper first estimates the initial scattering mapping of the global ambient light,and then refines the initial scattering mapping by introducing a high-order differential operator,thereby proposing a new variational model.In the process of solving the variational model,we apply the alternating direction multiplier method(ADMM).In order to verify the feasibility and effectiveness of the variational model based on higher-order differential operator proposed in this paper in scene restoration,this paper conducts rel-evant experiments on specific examples of sand and dust,underwater and haze scenes,and compares the results with those based on ROP and ROP~+.The experimental results of scene restoration show that the new variational model proposed in this paper is feasible,and has achieved good results in both dust,water,and haze scenes.That is to say,the new varia-tional model proposed in this paper is suitable for scene restoration in various environments.Compared to the scene restoration results based on ROP and ROP~+,the high-quality images restored in this paper have more realistic colors,fewer artifacts,and more details,The image is clearer and the restored result is closer to the original features of the image.The experimental results show that the proposed variational model in this paper is comparable to advanced scene restoration methods.