Research On Video Restoration Algorithm Based On Interframe Correlation

With the rapid development and progress of digital information technology,digital video technology in machine vision,traffic monitoring were used to various aspects,due to many factors,the video frames contrast,vague,detail information is not obvious[1-2],and the fog is also a kind of damage of video effect.Therefore,research on the theory and algorithm of digital video restoration become a hot research topic.Digital video reparation technology fill the unknown data by reasonable and efficient algorithm to estimate to the video frames.Video repair is the single frame image frame by frame processing,while there is no correlation between frame and frame,leading to a video repair computation is large.This paper is mainly about reparation for the interframe correlation digital video restoration algorithm and defogging algorithm.The specific study is as follows:(1)Digital video restoration by weighted fusion of fast marching in wavelet domain and block translationImage and digital video restoration is the process of calculating and repairing the damaged areas in the image or video.In this paper,based on the fast marching method in a small area of the damaged are widely used in the process of digital image restoration,and contacted the relationship of front frame and rear frame in video,it can be applied to video repair binding algorithm of fast marching method based on wavelet transform and translation method.Firstly,by two extraction wavelet transform image decomposition,the low frequency part adopts FMM algorithm to repair damaged areas of the image;Secondly,define and calculate pixel difference of template image in the current frame and next frame in the video,then it will be corresponding portion of next frame translated to the current broken images.Finally,a weighted average of pixel-level fusion about FMM algorithm repair Results Figure and translation algorithm results,complete the transformation from digital image restoration to digital video restoration.The proposed algorithm verify the feasibility,efficiency and practicality,repairing the damaged area of the video is better.(2)Digital video restoration by wavelet fusion of BSCB and block translationDigital video repairing algorithm is based on digital image restoration algorithm,BSCB(Bertalmio-Sapiro-Caselles-Bellester)model as a typical representative of the digital image restoration algorithm,can be combined with block translation method to realize digital video reparation.First of all,using BSCB model repair the damaged part of the current frame,translation to the current frame damaged parts using the next frame's complete parts by block translation algorithm.And using wavelet weighted fusion to restore the image by the pre-two algorithms.Through perfect refactoring capability of the wavelet,choose the mother wavelet reasonably,image can be decompose into average image and detail image.They were the structure part and detail part of unfused video frame.The low frequency part is processed by the weighted average method,while high frequency part has been adopted the absolute value algorithm.Finally,inverse transform of the fused coefficients is implemented in order to obtain the final fusion image.(3)Weighted least squares methods and dark channel prior video dehazing algorithm based on guided filterBased on the study algorithm of single image is currently processing on single frame image frame dehazing,video images dehazing by frame.But in this process,on the one hand,it has a large time consuming based on single image,so it is difficult to meet the requirements of real-time.In order to solve this problem,this paper proposes a combination of weighted least squares and oriented filtering algorithm dark channel prior to defog video frames.Algorithm using weighted least squares and oriented filtering of refine edge features one of the frame rate of video transmission diagram,make it as all universal transmission of video frame diagram.Algorithm increases the simply and effectively automatic levels adjustment steps to enhance the brightness of the image adaptively.The experimental results show that the video frame after dehazing by weighted least squares and oriented filtering has better resolution and the strong robustness,all those show the effectiveness of the proposed algorithm.