Image Restoration Algorithm For Underwater Video Sequence

This thesis focus on using image enhance and reconstruct method to improve the quality of underwater scenes from a video sequence severely distorted by radiation noise and water turbulence in the nuclear power plant.Because when the nuclear power plant is working,the radiation particle's moving is high-speed,non-linear and random,the restoration of this distorted image is very difficult.and the image influenced by water turbulence may cause pixel deviation.In the case of an underwater image containing turbulence noise,the image obtained by camera are likely blur,jitter and pixel offset.Therefore,to recover degradation images caused by different noise,we need to introduce different mathematical methods,degradation models,estimation criteria,and different recovery methods.This paper introduces the development of the image restoration technology and the main algorithm of image restoration.By using the information of adjacent frame we invent two algorithm:(1)For denoise the radiation particle and maintain the quality of underwater image,a reconstruction algorithm which integrated sequence image compare with image filtering was proposed.At first,an iterative sequence image compare algorithm was used to find noise spot.Then we use image filtering algorithm to replace the noise spot to recover the planar underwater scenes.With the experimental results,it is found that,in terms of restoration accuracy,visual effects and computational efficiency,the suggested algorithm always significantly outperforms better than the method employing image filter,leading to state-of-the-art performance on the task of underwater image restoration from video sequences.(2)A correction algorithm for distortion sequences was proposed based on B-spline free form registration to overcome pixel deviation from water turbulence.Because the pixel deviation caused by turbulence is nonlinear,and the global motion caused by the imaging system exist in practice,the correction method divided into two parts,the global motion estimation and the local non-rigid registration.By analysis the theory of underwater image restoration and the impact of water turbulence,a physical model of underwater image degradation system is proposed.Through this model,the local optimal solution of the offset vector of image is solved.