Research On Fast Restoration Of Remote Sensing Image Based On Coded Exposure And Vibration Detection

The acquisition of high resolution image is very important for remote sensing. However, the flutter of the satellite platform causes the blurring of the acquired image. And it becomes worse with the increasing of the exposure time. In the processing of recovering, due to the ill-posed problem, simple methods can not obtain great recovered results, meanwhile the ideal methods would consume a large amount of time.Coded exposure is put forward based on technology of spread spectrum. Taking advantage of the coded exposure, inverse filtering can be used directly in imaging restoration, since the problem of noise amplification caused by zero values of the spectrum is inexistent. In this article, we have proposed a fast restoration method for blurred vibration image based on coded exposure and vibration detection. The basic theory of coded exposure is researched, and the feasibility of the proposed method is verified. Both of the formula derivation of coded exposure and the mathematical model are also given. Due to that the form of code sequence would directly influence the recovered result, this article researches the choice of code sequence, and improves the criterion of code sequence based on traditional criterion of code sequence by taking into account factors such as signal-to-noise ratio and so on. The experimental results show that the code sequences chosen by the improved criterion have better recovering effect than that chosen by the tradition criterion. Considering that the code sequence is binary, we search the code sequence by genetic algorithm. With this algorithm, the search time is independent with the length of code sequence, and the optimal code sequence can be obtained quickly.This article has researched the real time vibration detection based on biaxial fiber-optic gyroscope and the accuracy of the detected data is then verified. The blur kernel is estimated with the detected vibration data by the statistic method, which changes the complex geometric calculation to the simple statistic calculation. Therefore, the algorithm complexity decreases. With the method mentioned above, the quick and exact PSF estimation method is obtained.System framework of coded exposure and vibration detection is constructed in this article, and the experimental facility is built. The simulation program is also written to obtain the experimental results, different recovery methods are compared with our method and different vibration conditions are considered. What's more, the experiment platform is built to do actual experiments, and the result is compared in both the subjective and objective aspect. The experimental results show that the proposed method can recover the blurred image efficiently and quickly. This efficient method is significant for acquiring the high-resolution remote sensing image quickly.