Research On Restoration Algorithm Of Multi Frame Turbulence-degraded Images

In the medium and long distance imaging system,atmospheric turbulence caused by temperature,wind speed and humidity distorts the propagation of light wave in the air,resulting in image quality degradation,such as geometric deformation and blurring.It is one of the key proble ms to effectively recovering the original target image from the degraded turbulence image.The effect of image restoration depends directly on the image processing technology.Therefore,it is crucial to study a fast and high-performance turbulence degradation image restoration algorithm,especially for online observation systems.Focusing on this issue,this paper mainly studies the image processing algorithm of atmospheric disturbance and turbulence degradation.The main work done is as follows:1.A method of multi frame registration turbulence image fusion based on adaptive guided filtering(AGFF)is proposed.Firstly,using the non-rigid image registration technology to eliminate the geometric deformation of the turbulent image.Then,the guidance filter and the high-frequency information of the mean image are used to construct the fusion weight map of each registration image.The fusion result of multi frame turbulence image can extract effective information and suppress noise and ghost phenomenon in degraded image.2.Aiming at the problem of uncertainty of feature decomposition of principal component analysis method,an improved principal component analysis(i-PCA)based multi frame turbulence-degraded image restoration method is proposed.Firstly,the multi-frame turbulence images are divided into several sub-image sets which are shift-invariant blurs.Then use the principal component analysis and maximum image similarity feature to obtain the restored sub-image for each sub-image set.Finally,all the recovered sub-images are recombined into one image.Compared with the effective filter flow(EFF)algorithm and the near-diffraction limited image reconstruction(NDL)algorithm,the results show that the proposed algorithm has high performance.3.Completing the indoor simulation turbulence introduced by temperature difference and outdoor near-ground turbulence image acquisition experiment.The proposed AGFF and i-PCA algorithms are used for processing and analysis,and the results demonstrate the practicability of the proposed algorithm for near ground turbulence image restoration.This research result will be applied to the online restoration of the image of turbulence degradation of crack defects of dam at medium and long distance(100m),and provide algorithmic support for the real-time online monitoring system of health of large hydraulic dam.