基于最优路径的无人机影像辐射区域网平差

无人机遥感能够提供高时空分辨率的影像数据，拥有广泛的应用前景。辐射校正将传感器记录的数据转化为地表反射率，是无人机数据定量化应用的前提。然而无人机数据易受光照等因素的影响，导致影像间存在不同程度的辐射差异，为多张无人机影像的辐射校正带来困难。基于影像间重叠区域的信息，辐射区域网平差能够获得全局最优的辐射校正参数，降低影像间的辐射差异，因此在实现影像的辐射校正方面具有巨大潜力。但大量待求解未知参数降低了辐射校正模型的求解效率，特别是在数据量急剧增加时，该问题更为突出。基于影像重叠区辐射信息建立的最优路径很好地考虑了影像间的辐射转化关系，可有效控制误差累计，为减少辐射区域网平差中未知参数的数量提供了一种思路。因此，本文将最优路径与辐射区域网平差相结合，以降低待求解参数的数量，在保证辐射校正精度的同时提高辐射校正模型求解效率，进而提升辐射区域网平差在大数据集上的应用潜力。

Optimal path-based radiometric block adjustment for unmanned aerial vehicle images

Unmanned aerial vehicle (UAV) remote sensing can provide imagery data with high spatio-temporal resolution, and has a wide range of application prospects. Radiometric correction converts the data recorded by sensors to ground reflectance and is the prerequisite for the quantitative application of UAV data. However, UAV data is susceptible to many factors such as illumination conditions, resulting in different degrees of radiometric differences between images, which makes it difficult for the radiometric correction of multiple UAV images. Based on the information of images' overlapping area, radiometric block adjustment can obtain the global optimal correction parameters, which has great potential in reducing the radiometric differences between the images and realizing the radiometric correction of images. But a large number of unknown parameters to be solved reduces the solution efficiency of the radiometric correction model, especially when the amount of data increases sharply, this problem will become more prominent. The optimal path established based on the radiometric information of images overlapping area takes into account the radiometric conversion relationship between the images, can effectively control the error accumulation, and provide an idea for reducing the number of unknowns in the radiometric block adjustment. Therefore, this paper combines the radiometric block adjustment with the optimal path to reduce the number of unknowns, and improve the calculation efficiency of the correction model while ensuring the accuracy of radiometric correction, thereby improving the application potential of the radiometric block adjustment on large datasets.