多视场遥感图像虚拟焦面拼接理论误差分析

Error Analysis of Remote Sensing Image Mosaic Based on Virtual Focal Plane

随着遥感对地观测技术的发展,对大视场、宽幅数据的需求日益增加,经常需要通过多幅遥感图像拼接来满足图像幅宽的要求。由多台相机同轨同时成像并拼接获取大幅宽遥感图像可以解决图像幅宽的问题。目前主要采用基于匹配的图像域拼接方法或根据成像几何关系生成虚拟拼接图像的方法获取宽幅图像。而生成虚拟拼接图像的方法物理意义明确且拼接后图像具有经典成像几何关系,成为图像拼接发展的趋势。文章根据同轨多视场图像虚拟焦面拼接原理,由摄影测量严格共线方程几何定位模型推导了多台相机不同视场图像的理论拼接误差计算公式,得出拼接误差的主要误差源;并仿真实验分析了拼接理论误差对图像拼接的影响,为多台相机拼接获取宽幅图像的设计和应用提供一定参考。

AbstractWith the development of remote sensing technology for the earth observation, the demand of wide-field images is increasing. Wide-field images are usually obtained by image mosaic, especially the multi-FOV remote sensing images of different sensors in the same satellite. Image matching and virtual mosaic image are the main methods to generate mosaic images. As virtual mosaic image has imaging geometry model with clear physical meanings, it is the future direction of image mosaic. In this paper, mosaic errors are first modeled by photogrammetry collinearity equation model according to the image mosaic algorithm based on virtual focal plane, and the main error sources of image mosaic are obtained. The effect of mosaic error to image mosaic is analyzed through simulation experiment, and the results can be used as the references for multi-FOV images mosaic designing and application.