无人机遥感已逐渐成为自然资源调查的重要技术手段，无人机倾斜摄影技术作为一种新型的测绘手段在基础测绘建设中发挥着重要作用。传统的垂直摄影测量由于受无人机姿态不稳定，数码相机畸变自校正精度不高、像点位移较大、区域网空三加密强度较弱的影响，需要布设大量的外业像控点来满足大比例地形图测量的需求，进而加大了外业工作成本、延长作业周期。

本文旨在优化无人机倾斜摄影测量像控点的布设，以期在保证测量精度的前提下减少外业像控点的数量。因此，本文主要针对像控点布设的网形和数量两方面开展优化研究。在像控点网形布设方面，本文设计并测试了4种布设方法，分别是四周法、中心增强四周法、品字形法和矩形格网法；在像控点数量优化方面，本文以矩形网布设为基础，根据基线数和航线跨度来设定参与区域网平差的基本定向点数量，共设计并测试了7种像控点数量布设方案。

选择河北省霸州市胜芳镇城区东部面积约1 km²的平原区域作为实验区，利用固定翼无人机对实验区进行倾斜摄影获取4个倾斜角度和1个垂直角度的地面影像数据，外业按照传统摄影测量外业规范要求测量像片控制点，然后针对像控点网形布设和数量布设方案分别进行研究，根据每种方案的空中三角测量（以下简称“空三”）精度分析提出像控点布设最佳的方案，同时对比倾斜摄影和传统垂直摄影两种摄影方式在布设方法和精度方面的差异，从而为采用倾斜摄影测量生产1:500平原地区大比例尺地形图提供理论依据和实践指导。研究结果和结论如下：

（1）像控点网形布设实验表明，在倾斜摄影像控点各种布设网形中，矩形格网法空三精度最优，其在像控点数量小于15的实验中检查点平面中误差为0.098 m、高程中误差为0.134 m，在像控点数量大于15的实验中检查点平面中误差为0.093 m、高程中误差为0.112 m；品字形次之，四周法最差，但4种布设网形的平面和高程精度差异不大，区域连接网稳定，在实际生产作业中可根据地形特点选择合适的布设网形。传统垂直摄影的各网形布设方案空三平面精度比倾斜摄影方式低0.03~0.04 m，高程精度低0.01~0.1m，垂直摄影最佳网形为中心增强四周法，各网形在高程精度方面差异较大，区域连接网不稳定，实际作业中可调整性较差。

（2）像控点数量布设实验表明，在7种像控点数量布设方案（数量分别为4、6、9、12、16、21、28）中，倾斜摄影方式随着参与空三平差的像控点数量增加，检查点平面中误差从0.161 m逐渐减小到0.083 m，高程中误差从0.475 m逐渐减小到0.081 m；在参与平差的像控点数量为12个时，空三精度已能满足1:500规范要求的检查点中误差平面和高程限差不超过0.175 m、0.15 m，实验中空三的精度随着像控点数量的增加而提高，但提高幅度逐渐较低。实际作业中，为了避免外业测量像控点工作量冗余，倾斜摄影方式布设像控时，建议相邻像控点之间的距离控制在400米以内即可。而传统垂直摄影方式在布设6个像控点开始，检查点平面中误差开始小于限差，而7种方案的高程中误差全部超过限差，无法满足规范的精度要求，实际作业中需要全野外布设高程点，外业工作量接近倾斜摄影的5倍。

本文通过对像控点布设网形和数量的研究，提出了在保证测量精度的前提下节省外业像控点测量的最佳解决方案，可为实际生产作业提供参考依据，减少外业生产成本，缩短作业周期，降低外业产生的风险，从而为基于无人机遥感的自然资源精细化调查提供技术支撑。

Unmanned aerial vehicle (UAV) remote sensing has gradually become an important technical means for natural resource investigation, and UAV oblique photography technology, as a new type of surveying and mapping method, plays an important role in basic surveying and mapping construction. Traditional vertical photogrammetry is affected by the unstable attitude of unmanned aerial vehicles, low distortion self correction accuracy of digital cameras, large pixel displacement, and weak encryption strength of aerial triangulation in regional networks. It requires the deployment of a large number of field ground control points (GCP) to meet the needs of large-scale topographic map measurement, thereby increasing the cost of field work and extending the work cycle.

This article aims to optimize the layout of GCP for unmanned aerial vehicle oblique photogrammetry, in order to reduce the number of field GCP while ensuring measurement accuracy. Therefore, this article mainly focuses on optimizing the layout and quantity of GCP. In terms of GCP network layout, this article designed and tested four layout methods, namely four perimeter method, center enhanced four perimeter method, chevron shape method, and rectangular grid method; In terms of optimizing the number of GCP, this article is based on the layout of a rectangular network, and set the number of basic orientationl points participating in block adjustment based on the number of photo base and flight route spans. A total of 7 layout plans for the number of GCP were designed and tested.

Selecting flat country of approximately 1 km² in the eastern part of Shengfang Town, Bazhou City, Hebei Province as the experimental area, a fixed wing drone was used to oblique photography of the experimental area to obtain ground image data of 4 oblique angles and 1 vertical angle. The field industry measured GCP according to traditional photogrammetric field standards, and then studied the network layout and quantity layout schemes of GCP, Based on the accuracy analysis of aerial triangulation for each scheme, the optimal scheme for the layout of GCP is proposed. At the same time, the differences in layout methods and accuracy between oblique photography and traditional vertical photography are compared, providing theoretical basis and practical guidance for the production of 1:500 large-scale topographic maps using oblique photography. The research results and conclusions are as follows:

(1) The experiment on the layout of GCP shows that the rectangular grid method has the best spatial accuracy among various network configurations of oblique photography GCP. In experiments with less than 15 GCP, the error in the check point horizonta is 0.098 m and the mean square error in vertical is 0.134 m, while in experiments with more than 15 GCP, the error in the check point horizonta is 0.093 m and the mean square error in vertical is 0.112 m; The product shape takes the second place, and the surrounding method is the worst. However, there is little difference in the accuracy of the horizonta and vertical of the four layout network shapes, and the regional connection network is stable. In actual production operations, suitable layout network shapes can be selected based on the terrain characteristics. The aerial triangulation horizonta accuracy of traditional vertical photography's various network layout schemes is 0.03-0.04 m lower than that of oblique photography, and the vertical accuracy is 0.01-0.1 m lower. The best network shape for vertical photography is the center enhanced four perimeter method. There are significant differences in vertical accuracy among different network shapes, and the regional connection network is unstable, resulting in poor adjustability in practical operations.

(2) The experiment of setting up the number of GCP showed that in seven different arrangement schemes (4, 6, 9, 12, 16, 21, 28, respectively), the oblique photography method gradually reduced the error in the check point horizonta from 0.161 m to 0.083 m, and the error in vertical from 0.475 m to 0.081 m as the number of GCP involved in spatial adjustment increased; When the number of GCP participating in the regional network adjustment is 12, the accuracy of aerial triangulation can meet the requirements of the 1:500 specification for check points with error plane and elevation limits not exceeding 0.175 m and 0.15 m. In the experiment, the accuracy of aerial triangulation increases with the increase of the number of image control points, but the improvement gradually decreases. In practical work, in order to avoid redundant workload of field measurement GCP, it is recommended to control the distance between adjacent GCP within 400 meters when arranging GCP using oblique photography. However, the traditional vertical photography method starts with the placement of 6 GCP, and the error in the horizonta of the check points starts to be less than the limit. However, the vertical error in all 7 schemes exceeds the limit, which cannot meet the accuracy requirements of the specifications. In actual operations, it is necessary to arrange vertical points throughout the field, and the field workload is close to 5 times that of oblique photography.

This article proposes the best solution to save field GCP measurement while ensuring measurement accuracy by studying the layout and quantity of GCP. It can provide reference for actual production operations, reduce field production costs, shorten work cycles, and reduce risks generated by field operations. This provides technical support for the refined investigation of natural resources based on UAV sensing.