弱交会条件下顾及同轨约束关系的卫星影像区域网平差方法研究

As a basis of subsequent geometry processing for satellite images, block adjustment has been widely used in satellite photogrammetry. However, there are two problems of conventional block adjustment for satellite images in weak convergence condition. First the block adjustment will be abnormally solved which also leads to the adjustment process cannot be converged. Second the strip constraint condition is not considered, so more ground control points are needed for block adjustment. Therefore the research group intended to propose a block adjustment method for satellite image based on strip constraint with weak convergence due to the block adjustment theory. By the theoretical analysis of different weak convergence conditions the threshold will be determined. The mathematic model and solving method of planar block adjustment with weak convergence will be researched and the strip constraint condition is considered. The generation method of virtual strip image and its RFM will be researched. Compensation grid model is used to reduce the fitting error of virtual strip image's RFM. Under the analysis of the geometric relationships between the virtual strip image and the single image, the orientation parameters of single images will be considered. In the practical applications, this research can provide the image orientation parameters for large regional ortho-rectification. It reduces the number of ground control points required by the ortho-rectification and guarantees DOMs of the entire region in a unified space reference.It also ensures the edge matching accuracy between the adjacent DOMs. It provides security for collecting and updating the fundamental geographic information and various emergency surveying and mapping.

区域网平差作为卫星影像后续几何处理的关键技术，在航天摄影测量领域得到了广泛应用。然而在弱交会条件下，若采用常规区域网平差方法会带来两个问题①平差求解异常或平差过程不收敛。②没有考虑同轨影像间的约束关系，因而平差所需地面控制点较多。因此，本研究基于经典区域网平差理论，拟提出一种弱交会条件下顾及同轨约束关系的卫星影像区域网平差方法，理论分析不同情况弱交会条件阈值的确定，以及在该条件下卫星影像上区域网平差的数学模型及求解策略。在此基础上充分顾及同轨约束条件，研究虚拟条带影像的还原方法，利用补偿格网模型降低虚拟条带影像RFM的拟合误差。通过条带影像与单景影像的几何对应关系，确定最终单景影像的定向参数。本研究可为大区域卫星影像正射纠正服务，减少地面控制点的数量，保证纠正后卫星影像在统一的空间基准下，保障相邻影像之间的接边精度以及绝对定位精度，为基础地理信息的快速获取、更新和各种应急测绘提供技术保障。

项目针对卫星影像区域网平差过程出现弱交会条件下的两个问题①平差求解异常或平差过程不收敛，②没有考虑同轨影像间的约束关系，平差所需地面控制点较多，因此本研究基于经典区域网平差理论，提出一种弱交会条件下顾及同轨约束关系的卫星影像区域网平差方法，理论分析不同情况弱交会条件阈值的确定，构建了卫星影像上区域网平差的数学模型以及设计了求解策略。在此基础上充分顾及同轨约束条件，研究虚拟条带影像的还原方法，利用补偿格网模型降低虚拟条带影像RFM的拟合误差。通过条带影像与单景影像的几何对应关系，确定最终单景影像的定向参数。此外，通过本研究进一步发现，SAR卫星影像的大区域正射纠正同样可以采用本项目方法获得较好的精度。利用本项目的研究成果，针对江西省范围的资源三号光学卫星影像，进行了区域网平差试验，稀疏控制条件下检查点平面中误差3m，高程中误差2m。针对全国范围覆盖的高分三号SAR影像10m分辨率，无控制条件下检查点中误差8m，且连接点精度均在1个像素左右。本研究可为大区域光学/SAR卫星影像正射纠正和DSM生产服务，减少地面控制点的数量，保证纠正后卫星影像在统一的空间基准下，保障相邻影像之间的接边精度以及绝对定位精度，为基础地理信息的快速获取、更新和各种应急测绘提供技术保障。