Restoration Of Irregularly Sampled Optical Remote Sensing Images

TDI-CCD is widely applied in passive optical remote sensing due to the highspatial resolution and radiation responsivity it can achieve. Satellite vibrations,however, causes irregular geometrical distortion in the corresponding images.Although the effect is trivial in low resolution images, it becomes more noticeable asthe number of TDI line and the spatial resolution of imaging system go higher.The effects of vibrations can be reduced by either suppress vibrations directly orrestore the degraded images by certain methods. As the control precision of satelliteplatform is currently limited, a new restoration method for irregularly sampled opticalremote sensing images, based on the study of satellite vibration and the study ofoptical remote sensing imaging model, is constructed in solving the practical problemfacing by high resolution imaging. Image deconvolution and denoising are alsoinvolved in the method. Finally, simulation results are demonstrated to proverestoration quality and efficiency of the method.The thesis is arranged as following:1. Satellite vibration and its effects. Based on the analysis of vibration sources,modes and frequency distribution, satellite vibration model is constructed and themethods for acquiring vibration parameters are discussed. After that, image shiftcaused by vibration is studied on the basis of time delay integration theory.2. Imaging model of irregularly sampled optical remote sensing images.According to the different between degradation phenomena, the effects of vibrationare classified and modeled. Theoretical analysis and simulation on these models arethen applied to study the specific effects of different vibrations. Eventually,incorporating with the discussion of the other degradation factors in remote sensingimaging chain, an overall imaging model is constructed.3. Restoration of irregularly sampled optical remote sensing images. Themathematical problem of distortion correction is proposed at the first stage forirregular geometrical distortion. On the basis of this, blur and noise are introducedinto the problem so that the overall problem of image restoration is acquired. Finally,by the application of certain mathematical tools, the restoration problem is solved inan effective way.4. Simulation and restoration quality evaluation. Due to the complexity ofirregularly sampled optical remote sensing images, multi-criteria strategy is appliedfor restoration quality evaluation. Simulation results of restoration are evaluated viathe criteria and overall performance of the restoration algorithm is concluded at last.