Research On The Key Technology Of Atmospheric Degraded Remote Sensing Image Restoration

With the rapid development of high resolution astronomical imaging and earth observation imaging technology,atmospheric turbulence has gradually become the bottleneck which restricts the improvement of the optical imaging system performance.Therefore, the demand for removal of atmospheric disturbances becomes more and more urgent. When the optical imaging system through the atmosphere imaging on the surface or dimensional interested target, it will inevitably affected by atmospheric disturbances.Aiming at this problem,based on the conclusion of existing research results of atmospheric degradation image restoration algorithm, with the support of national natural science fund project,this dissertation commits a deep research and discussion on the high definition recovery and reconstruction of the image of atmospheric degradation problem. Introducing astronomic mesoscale meteorological model and a single star in the field of astronomy SCIDAR observation technology for atmospheric turbulence profile intensity, it can estimate atmospheric MTF,which is described as the priori information in the imaging moment and incorporated into image restoration model, which can eliminate the atmosphere’s blur in observed image. In this dissertation, the major research content and innovations include:1.The study about the characteristics of atmospheric turbulence and its influence on the optical imaging system is studied and quantitatively analyzed. Firstly,it detail introduces the formation mechanism of atmospheric turbulence and the parameterized description model; Secondly, based on the theory of atmospheric turbulence transmission,ground-based or air-based optical imaging system degradation effect on turbulence has been quantitatively analyzed respectively from the atmospheric coherence length and the angle of arrival/fluctuation. When atmospheric turbulence is existed, ground-based telescope angular resolution imaging depends on the coherence length of the optical system of the atmosphere rather than the aperture of the optical system. The affection of ground resolution in the ground-based optical remote sensing space system is about 2cm～10cm;2.The atmospheric turbulence profile acquisition technology is researched. Atmospheric optical refractive index structure constant is a significant physical parameter of describing atmospheric optical turbulence characteristics.The study on atmospheric turbulence profile has been researched in this paper respectively from two aspects, the model calculation and the measurement method. Model calculation method uses mesoscale meteorological model calculation and forecast of meteorological parameters in the imaging moment and bases on the atmospheric optical turbulence parametric model to calculate atmospheric optical refractive index structure constant. Single Star SCIDAR technology utilizes the optical wave propagating and the spatial correlation properties for self-correlation and cross-correlation, and it obtains the atmospheric turbulence profile intensity distribution through the simulated annealing algorithm inversion. The atmospheric turbulence MTF can be directly estimated according to atmospheric turbulence profile calculated or actual measured by the model;3.Imaging time overall atmosphere MTF estimation algorithm is designed. Overall atmospheric MTF is mainly composed of the product of atmospheric turbulence MTF and aerosol MTF, which respectively can be described by the meteorological parameters Cn2 and aerosol size distribution. According to the imaging time recorded meteorological data in the same area and using the two parameters, the estimation for atmospheric turbulence MTF,aerosol MTF and overall atmospheric MTF can be obtained.Take use of atmospheric MTF into the restoration model, which can improved the performance of restore algorithm;4.Modified wiener filter and IBD algorithm combined with overall atmospheric MTF estimate are proposed. Though the optimization of incremental wiener filtering and degradation function of IBD algorithm, according to the estimation for overall atmosphere MTF, it can obtain the improved algorithm which includes atmospheric conditions prior information in imaging time, which can more accurately describe the atmospheric degradation in imaging time, and be advantageous to the atmospheric degradation image restoration. Experimental results show that, atmospheric MTF can accurately reflect the imaging effect caused by atmospheric turbulence; hence it can be treated as reliable prior information of eliminating atmospheric blur;5.APEX algorithm combined with adaptive parameter and atmospheric MTF estimation are proposed.Aimed at the limitations of APEX algorithm,this dissertation puts forward a method for weighted adaptive parameter determination, meanwhile,adding a Single Star SCIDAR measured profile data estimation for atmospheric turbulence MTF to modify the data of original optical transfer function, which makes the algorithm can contain a priori information reflects imaging time atmospheric condition. The experimental results show that the improved algorithm can better describe the overall characteristics of image spectrum information and the quality of reconstructed image can be significantly improved;6.The maximum likelihood algorithm combined with noise removing is proposed. Firstly, use BM3D to removal the image noise, which can reduce the noise effect on the stability of the algorithm; Secondly, EM algorithm is used to estimate the parameter of noise, which can improved the stability of the algorithm;Finally, in the iteration process it adds the dynamic updated PSF iterative formula, which reduce dependence on PSF initial value of the restoration results. The experimental results show that, the algorithm mentioned effectively overcome noise amplification and dependence on initial estimate PSF in the recovery process, which also improves the quality of the restored image.This method can be used for high-resolution reconstruction of degraded images with noise;7.The problem of removing atmospheric haze based on dark colors priori is researched. For He Kaiming proposed in his paper that the algorithm efficiency is low and the defect that image bright region cannot meet the needs of dark channel prior, it has been proposed that the use of optimized guide filter transmittance estimated can improve the bright regional dark colors priori failure problem. Finally multi-resolution computing can further improve the efficiency of the algorithm. Experimental results show that the improved algorithms ensure that the basis of image quality on defogging and correct the problems that image bright areas tonal distortion in the process of eliminating the fog. Meanwhile, the efficiency of the algorithm has been significantly improved and broadens the application ranges of dark colors defogging algorithms.