Research On Super-resolution Reconstruction Of High Resolution Video Satellite Images

Unlike traditional satellite static imaging,video satellites can image the same scene continuously.In view of this feature,different information between frames can be extracted to improve the image quality.Therefore,in order to improve the image resolution,this paper studies the application of super-resolution reconstruction.Super resolution reconstruction is a method to improve image resolution through computer methods combined with rigorous mathematical physics methods and theoretical simulations when the resolution is limited and the original hardware is not changed.The method has the advantages of higher flexibility than hardware,shorter development cycle,lower cost,and higher feasibility.The high-resolution images obtained have a good image visual effect,and have broad commercial,military and other application prospects,such as target tracking,detection,identification and so on.This paper focuses on video satellite imagery and studies a video satellite visualization system based on super-resolution reconstruction algorithm.The main contents are as follows:1.A video satellite staring degraded model was proposed.This paper proposes to analyze the impact of degraded models on super-resolution reconstruction from different perspectives of “active viewpoint” and “passive perspective”.Taking video satellite remote sensing images as the research object,this paper adopts active viewpoint staring degraded imaging model.Using the camera stance as a reference system effectively avoids the poor information utilization of and artificial errors in the traditional passive view.Its establishment provides a theoretical model and solution for the super-resolution algorithm and occupies an important position in the entire algorithm.2.The intermediate convex sets mapping(ICSM)algorithm for multi-frame superresolution reconstruction is proposed.This algorithm is based on the active viewpoint degradation model.After the implement of SIFT algorithm-based image registration and blur kernel estimation steps,ICSM algorithm is used to improve the image resolution.The algorithm establishes a constrained set based on the intermediate degrading process and a corresponding point projection operator.The point projection operator corrects the gray value of the high-resolution image frame by frame,and finally the intersection of the reconstructed high-resolution image constrained convex set.The proposed reconstruction makes the model of super-resolution images more complete and effective.3.The final selection of the article evaluate the reconstructed image with nonreference image quality evaluation indicators.This evaluation completes the superresolution reconstruction system.Based on these indicators,analogy of multiple sets of comparative experiments and a large number of experimental data to verify the validity of the algorithm.4.An MTFC algorithm for single-frame image super-resolution image is proposed.The on-orbit measurement of the satellite image modulation transfer function is realized by the improved inclined-edged edge method proposed in this paper.Wiener filtering is used to achieve the modulation transfer function compensation and single-frame image super-resolution reconstruction.In the end,both real-time satellite in-orbit images are monitored.The role of image quality can also play a role in the modulation transfer function compensation and resolution improvement.In summary,this paper developed a complete super-resolution reconstruction system.This system has also developed related functions such as modulation transfer function measurement and compensation,image restoration,image registration,image extraction in the same area,and image quality evaluation.The establishment of this system has important theoretical innovation and application value for improving the resolution of the video satellite images.