大规模空中三角测量的智能分区与合并算法

Intelligent partitioning and merging of large-scale aerial triangulation

针对空中三角测量的处理时间随图像数量的增加呈指数级增长,处理大规模数据集需要大量时间的问题,该文提出了一种测区智能分区与合并算法。该算法可以自动地将无序的图像集分割成若干个有重叠的子集,将每个子集可以并行处理,根据重叠图像的连接点,将各个子集的重建结果合并在一起。实验结果表明,该算法不仅保证了时效性,同时流程也简单,处理速度很快。在一个有10处理节点的集群系统中,该算法成功地处理了大规模航空影像数据集,重建的时间效率和精度满足实际生产要求。与不分区进行比较,时间可以节省至少一半。

To solve the problem that the processing time of aerial triangulation increases exponentially with the number of images and a lot of time is spent to process large scale data,this paper proposed an intelligent partition and merging algorithm.The algorithm could automatically divide the unordered image and set into several overlapping subsets,and then each subset could be processed in parallel.Finally,according to the connection points of overlapping images,the reconstruction results of each subset were merged together.Experimental results showed that the algorithm was not only effective but also simple and fast.We had successfully processed large-scale aerial image datasets in a cluster system with 10 processing nodes.The time efficiency and accuracy of reconstruction meet the needs of practical production.Compared with no partition,time could be saved by at least half.