提出一种基于虚拟大相机的光学遥感卫星多相机系统成像数据的拼接方法,实现多相机影像的高精度拼接。该方法根据各单相机的几何成像模型,构建虚拟大相机的几何成像模型;利用坐标正算、反算过程对各单相机影像进行间接法几何纠正,得到虚拟大相机图像坐标系下的各虚拟单相机影像;将各虚拟单相机影像基于影像坐标信息拼接处理后得到最终的虚拟大相机影像。该方法利用虚拟大相机实现多相机影像的高质量拼接,为后期处理提供了与之相对应的高精度有理函数模型,可用于不同数量、不同分辨率的多相机系统实现全自动智能地面预处理。

A high accuracy image mosaicking method based on big virtual camera is proposed for multi-camera system on the optical remote sensing satellite to realize the high-precision stitching of multi-camera images. A big virtual camera is built according to the geometric imaging model of each single camera, and the geometric correction is performed by coordinate forward and back calculation based on their imaging models to obtain each single virtual image in the virtual camera image coordinate system. The final stitched image can be obtained by stitching single virtual image based on its coordinate information. By using the concept of big virtual camera, the high accuracy stitched image and corresponding rational function model for post processing are obtained at the same time. It can be used for multi-camera system with different quantities and resolutions to realize automatic intelligent ground preprocessing.