Image Restoration Method And Vibration Degradation Simulation Of TDI Remote Sensing Image

In recent years,remote sensing technology has been widely used in such fields as resource investigation,topographic investigation and military reconnaissance,occupying a noticeable place in national security and national economy.Space cameras,as the "eyes" of remote sensing satellites,play a crucial role in obtaining clear and reliable remote sensing data.According to different sensor types,space cameras can be divided into array cameras and integration cameras.Time delay integration(TDI-CCD)cameras are widely used in the field of remote sensing because of their high imaging quality.However,due to the influence of the space environment,the TDI-CCD remote sensing camera may suffer from vibrations during the imaging process,resulting in blurred images.In this paper,TDI-CCD remote sensing simulation model is constructed based on the imaging mode and degradation mechanism of TDI-CCD remote sensing camera.The validity of the simulation model is proved by the analysis of degradation rules of remote sensing images under different integral series and vabration frequency.In the optimization method based on similarity degree of inter-row kernels,this paper first analyses the influence of image block length and similarity degree of inter-row kernels on the restoration of TDI remote sensing images which suffer from vibrations.It is found that with vibration components along scanning direction,the increase of image block length contributes to restoration effect of vibration images with high similarity of inter-row kernels while ruins the recovery effect with low similarity of inter-row kernels.A relationship formula of image block length and similarity degree of inter-row kernels is built according to the above conclusion.Moreover,an optimization method is proposed by adaptive adjustment of image block length with similarity degree of inter-row kernels.Feasibility of the optimization method proposed is proved by comparing the recovery effect of TDI vibration images when the image block length is fixed and adaptive adjusted.In the TDI remote sensing recovery method based on optimization of numerical fidelity term,this paper first analyses the causes of ringing artifacts in the recovery process,which shows that the ringing artifacts are mainly caused by the inaccurate estimation of the fuzzy kernel and improper image model.In the imaging process with vibration,fuzzy kernels cannot achieve high precision since inter-row kernels are inconsistent.Thus the second order information of the image is introduced on the basis of total variation method to construct a more accurate image model and improve the recovery effect.To demonstrate the universality of the new algorithm under different vibration conditions,analysis of the recovery effect of TDI remote sensing images which are respectively recovered by Richardson-Lucy(RL)algorithm,total variational method and the new algorithm in different integral series and vibration frequency is presented.Finally,several real remote sensing images with vibrations in two-dimensional direction are recovered by the algorithm proposed in this paper,and an objective analysis of the recovery effect is presented by analyzing the structural similarity index(SSIM)and the peak signal noise ratio(PSNR),which proves the practicability of the proposed algorithm.