卫星视频影像超分辨率重建方法对比分析

Comparative Analysis of Super Resolution Reconstruction Method for Satellite Video Image

多帧影像的超分辨率重建是指利用影像之间互补信息提高影像分辨率的技术,在卫星视频领域可以充分利用影像的帧间互补信息,采用超分辨率重建技术提高视频的分辨率,能够在降低硬件开发成本的同时拓宽卫星视频的应用领域。但卫星视频是目前遥感领域的新型数据源,已有的传统重建方法是否适用于卫星视频影像,以及不同重建方法的适用性条件及效果还没有对比研究。因此,针对上述问题本文采用"吉林一号"卫星视频数据,对比分析了几种不同运动估计条件下6种重建算法的适用性和优缺点,并得出了在Vandewalle运动估计方法下稳健的超分辨率重建算法、IBP(Iterated Back Projection)算法、PG(Papoulis-Gerchberg)算法、POCS(Projection onto convex sets)算法的重建效果和适用条件,为视频卫星重建方法的选择提供了可参考的结论。

Super-resolution reconstruction of multi-frame images refers to the technology of using complementary information between images to improve image resolution.In the field of satellite video,you can make full use of the complementary information between the frames of the image,and use super-resolution reconstruction technology to improve the resolution of the video,which can expand the application field of satellite video while reducing the cost of hardware development.However,satellite video is a new data source in the field of remote sensing.Whether existing traditional reconstruction methods are suitable for satellite video images,and the applicability conditions and effects of different reconstruction methods have not been studied.Therefore,in view of the above problems,this article uses the satellite video data of"Jilin No.1",and comparatively analyzes the applicability,advantages and disadvantages of the six reconstruction algorithms under different motion estimation conditions.And the reconstruction effect and applicable conditions of the robust super-resolution reconstruction algorithm under the Vandewalle motion estimation method,IBP(Iterated Back Projection)algorithm,PG(Papoulis-Gerchberg)algorithm,and POCS(Projection onto convex sets)algorithm are obtained.The selection of video satellite reconstruction methods provides reference conclusions.