针对计算机视觉领域中的增量式运动恢复结构和摄影测量学中的POS辅助光束法平差的精度进行了比较研究。首先介绍了SFM的基本原理与主要流程,并从理论上分析了两种方法的异同;然后针对定位精度进行了对比研究,分别利用嵩山遥感定标场的无人直升机和有人驾驶飞机的航摄数据进行了精度对比试验,结果表明:在不依赖POS数据的情况下,增量式运动恢复结构可以达到与POS辅助光束法平差相当的精度,两种方法均可以满足1:500、1:1000成图要求;最后,针对不同作业需求给出了航摄数据后处理参考意见。
An accuracy comparison research was conducted between the incremental structure from motion in computer vision and the POS-aided bundle adjustment in photogrammetry. Firstly, the basic principle and workflow of SFM were introduced, and the similarities and differences between the two methods were analyzed theoretically. Then the positioning accuracy was compared, using the unmanned helicopter and manned aircraft aerial survey data in the Songshan remote sensing calibration field. Experimental results show that without POS data, the incremental structure from motion method can achieve equivalent accuracy to the POS-aided bundle adjustment. Both methods can meet the requirement of 1:500 and 1:1000 mapping. Finally, suggestion was proposed for aerial data post-processing according to different operational requirements.