Image Degradation Modeling Due To Satellite Platform Vibration And Blur Parameter Identification

Improving the quality of satellite images is one of the most critical tasks on the satellite applications. With the improvement of the satellite resolution, the satellite platform vibration has become one of the key factors affecting satellite image quality. This paper researches on the blurring problem of satellite image caused by vibration of the platform,and the method of how to determine the kernel function on the image restoration.Firstly, the mechanism of the satellite platform vibrations is analyzed and a model describing the impact of the satellite platform vibration with a delay integration CCD on imaging quality is established. It shows that the satellite platform vibration seriously affects the quality of satellite image, and the more series of the delay integration CCD, the severer impact on image quality; secondly, the motion-blur parameters are identified on space domain, frequency domain and cepstral domain, and on the cepstral domain of the differential image, respectively. Conclusions show that the four methods perform well on different case. the identification method on space domain by directional derivative is simple and intuitive but with poor results; Radon transform in the frequency domain and the cepstrum method reduce good and high-precision results, but both methods are susceptible to noise and fail in a large noise intensity; relative to the above three parameter identification methods, the new method on the cepstral domain of the differential image gets more high-accuracy results, and is robust to noise; but it could not get accurate recognition results on very small scale motion. Finally, the motion-blur image is restored with sparse regularization. Experimental results show that this method, which models the impact on the image by the platform vibration and identifies the blur parameters, gets much better restored image than blind image restoration method.