Research On Super Resolution Reconstruction Of Satellite Video Motion Scene

In recent years,satellite video as a new product in the field of remote sensing,so that people can get the ground video taken in space,to achieve a change from a single image to continuous dynamic image.However,satellite video image resolution is low due to external disturbances,imaging blur,atmospheric turbulence and other factors in the process of imaging.Therefore,it is very important to study how to improve the resolution of satellite video image by software using super-resolution reconstruction technology.It has very important research significance and application value.In this paper,the problems in the process of satellite video super-resolution reconstruction are studied.The main contents are as follows:1.Based on the satellite video,The thesis analyzes the imaging characteristics of the satellite video,the different between satellite video images and ordinary video images,remote sensing images.Aiming at the complex motion characteristics in satellite video scene,The thesis analyzes its impacts on the super-resolution reconstruction.2.Aiming at the static background and complex dynamic target coexistence in satellite video scene,the classical super-resolution reconstruction method has the problem of moving target tailing and blurring.This paper propose a satellite video super-resolution reconstruction method combining motion segmentation and optical flow estimation.Firstly,the edge information of the moving target is obtained by the motion segmentation method,and then the motion estimation of the pixel is carried out by the optical flow method to obtain the initial motion vector field.Finally,using the joint edge and motion vector field combined constraints,and unified motion vector in the moving target,and it is compensated to the iterative back projection super-resolution reconstruction model to obtain the final image.The experimental results show that the method can solve the problem of tailing of moving objects.3.Aiming at the unavoidable problem of motion estimation error,this paper proposes an adaptive robust regularization super-resolution reconstruction method.In the framework of regularized super-resolution reconstruction method,a robust estimation theory is introduced,Use the new robust estimation function to replace the fidelity item in the regularization model,and use the bilateral filter as the regular term,so that this method can reduce the contribution of low resolution images with motion estimation errors in the reconstruction process.The experiment of super-resolution reconstruction is carried out by using ordinary video and satellite video respectively.It is proved that this method has good robustness to motion estimation error.Finally,the method is developed in the super-resolution reconstruction system.