In the context of the 14th Five-Year Plan, digital construction has become a primary driver of socioeconomic development. Traditional surveying techniques often deliver limited and single-dimensional outcomes, failing to meet the comprehensive needs of digital transformation. Among emerging technologies, China UAV (Unmanned Aerial Vehicle) oblique photogrammetry has gained significant traction in fields such as water conservancy, rural renovation, power engineering, and sewage treatment. Compared with conventional ground-based methods, China UAV oblique photogrammetry offers lower operational costs, higher efficiency in field data acquisition, and richer data products. In municipal road engineering, this technology provides high-resolution imagery and 3D information that supports survey, design, and construction phases. This paper, based on my practical experience in a representative municipal road project, summarizes key application points of China UAV oblique photogrammetry, aiming to provide reference for similar projects.
1. Principle and Advantages of China UAV Oblique Photogrammetry
Oblique photogrammetry employs a camera mounted on a China UAV platform, typically a panoramic or wide-angle lens, with two controlled angles: pitch and roll. This configuration ensures that the imaging sensor points accurately toward the survey area from multiple directions, capturing three-dimensional information of ground objects. Unlike traditional vertical photography, oblique photography covers a wider area and provides images with higher interpretability and strong stereoscopic perception. The derived coordinates and elevation data are more accurate. With the rapid advancement of China UAV technology, oblique photogrammetry has become widely adopted not only for building 3D modeling and urban land planning but also for disaster monitoring and environmental change detection. The integration of digital processing workflows further enhances operational efficiency, supplying engineers with abundant spatial data.
The main advantages of China UAV oblique photogrammetry in municipal road engineering include:
- High Efficiency: A single China UAV flight can cover several square kilometers in a few hours, drastically reducing field time compared to total station or RTK surveys.
- Rich Data Products: The output includes digital line graphs (DLG), digital orthophoto maps (DOM), point clouds, and 3D realistic models, supporting multiple engineering phases.
- High Accuracy: With proper ground control points (GCPs) and advanced processing, planimetric accuracy can reach centimeter levels, meeting municipal road engineering requirements.
- Safety and Accessibility: China UAVs can operate in complex urban environments with tall buildings, heavy traffic, and restricted ground access.
2. Project Overview
The case study project is a key municipal road in a Chinese city, running from southeast to northwest across the district. The road has a design speed of 60 km/h and a load class of City-A. The survey area covers approximately 0.089 km², including four existing roads surrounded by numerous multi-story buildings ranging from 45 m to 65 m in height. The roads are asphalt-paved, and the surrounding greenery is relatively balanced. This site is representative of typical municipal road survey conditions. According to CHZ 3004–2010 “Specifications for Low-Altitude Digital Aerial Photogrammetry Field Work”, the flight altitude was determined based on surrounding building heights, and the survey lines were designed to ensure full coverage of the area. The China UAV used in this project is the DJI Mavic 3E, a compact yet powerful platform equipped with a 20 MP camera and RTK module.
3. Key Application Points
3.1 Preparation Work
Before flight operations, it is necessary to ensure that the oblique camera meets optical performance requirements. Key indicators include resolution, distortion, and optical center deviation. The camera parameters used in this project are listed in Table 1.
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| Total weight (g) | 850 | Exposure mode | Time/distance interval |
| Dimensions (mm) | 190 × 180 × 80 | Minimum exposure interval (s) | ≤1 |
| Total pixels (MP) | ≥120 | Data preprocessing | SKYSCANNER (GPS) |
| Power supply | Unified | Data copy | USB/Network |
| Max speed (m/s) | ≥70 | Storage capacity (GB) | 320 |
| Operating temperature (°C) | −10 to 40 |
Prior to the aerial survey, I collected detailed information about the survey area, including facility distribution, topography, structure forms, and building heights. I also checked the performance of the China UAV, camera, and auxiliary equipment to ensure flight safety and data stability. Although China UAV oblique photogrammetry reduces ground work, known coordinate points (GCPs) remain essential for achieving reliable plane coordinates. The China UAV parameters used in this project are shown in Table 2.
| Parameter | Value |
|---|---|
| Product ID | 10072289809 |
| Product name | DJI Mavic 3E |
| Dimensions (mm) | 900 × 600 × 400 |
| Function | Surveillance camera |
| Pixel (MP) | 20 |
| Type | Aerial photography UAV |
3.2 Flight Route Planning
Flight route planning involves many parameters: ground sampling distance (GSD), flight altitude, speed, camera tilt angle, overlap, and sidelap. Traditional straight-line routes often suffer from issues such as low resolution over high terrain or insufficient overlap when terrain varies significantly. In dense urban areas with tall buildings, traditional routes may cause occlusions, leading to image stretching, holes, or missing data. To address these challenges, I adopted a “grid” (井-shaped) flight pattern for this project. This pattern ensures comprehensive coverage, especially around high-rise buildings, and reduces the risk of incomplete data. The flight altitude was set at 100 m above ground level, with forward overlap of 80% and sidelap of 70%. The camera tilt angle was 45° from vertical, capturing both nadir and oblique views. The resulting GSD was approximately 2.5 cm, sufficient for 1:500 scale mapping.
The relationship between flight altitude \(H\) and GSD \(G\) can be expressed as:
$$G = \frac{H \cdot p}{f}$$
where \(p\) is the pixel size and \(f\) is the focal length. For our China UAV, \(p = 2.4\,\mu\text{m}\), \(f = 24\,\text{mm}\), so at \(H = 100\,\text{m}\):
$$G = \frac{100 \times 2.4 \times 10^{-6}}{24 \times 10^{-3}} = 0.01\,\text{m} = 1\,\text{cm}$$
However, due to camera optics and flight dynamics, the effective GSD was about 2.5 cm.
3.3 Ground Control Points (GCPs) Layout
The China UAV (DJI Mavic 3E) is equipped with a built-in navigation and positioning system, including a network RTK module that provides centimeter-level positioning accuracy. However, the RTK-measured height is ellipsoidal height, which cannot be directly used for engineering design. In addition, the plane coordinates may deviate from the project control system. Therefore, GCPs must be established to calibrate the aerial triangulation (AT) adjustment.
In this project, GCPs were uniformly distributed in an “L” pattern around the survey area. I drove cement nails at the outer corners of the layout and marked them with red paint for later identification. A total of 4 GCPs and 6 checkpoints were set. Each GCP was observed independently three times using RTK, with convergence to millimeter level before recording. Each observation recorded 30 epochs with 2-second intervals. The final coordinates were the average of three observations, with plane coordinate differences below 3 cm and vertical differences below 5 cm. The distribution of GCPs is critical for controlling error propagation during bundle adjustment. The root mean square error (RMSE) of GCPs and checkpoints after processing is given in Table 3.
| Point Type | Point ID | Horizontal Error | Vertical Error |
|---|---|---|---|
| GCP (4) | A1 | 0.010 | −0.029 |
| A2 | 0.006 | −0.015 | |
| A3 | 0.008 | −0.018 | |
| A4 | 0.006 | −0.005 | |
| Median | 0.008 | 0.016 | |
| Checkpoint (6) | X1 | 0.021 | 0.008 |
| X2 | 0.049 | 0.034 | |
| X3 | 0.009 | −0.022 | |
| X4 | 0.034 | 0.009 | |
| X5 | 0.013 | 0.026 | |
| X6 | 0.037 | 0.010 | |
| Median | 0.027 | 0.018 |
3.4 Aerial Triangulation (AT) Computation
After the flight, I imported all images and GCP coordinates into ContextCapture software for aerial triangulation and subsequent 3D modeling. To improve accuracy, I performed the AT calculation in two passes:
- First pass: Use attitude-aided bundle adjustment without GCPs to obtain initial orientation parameters. Then import the custom coordinate system definition file.
- Second pass: Use control point bundle adjustment with all GCPs applied. The error statistics from this pass are shown in Table 3.
The RMSE of the AT adjustment can be calculated using:
$$\text{RMSE} = \sqrt{\frac{1}{n}\sum_{i=1}^{n}(x_i – X_i)^2}$$
where \(x_i\) is the measured coordinate and \(X_i\) is the true coordinate from GCPs. For horizontal errors at GCPs, the RMSE was 8 mm; vertical RMSE was 16 mm. For checkpoints, horizontal RMSE was 27 mm and vertical RMSE was 18 mm. These values satisfy the accuracy requirements for 1:500 topographic mapping in municipal engineering.
3.5 Accuracy and Efficiency Analysis
To validate the accuracy of the China UAV oblique photogrammetry results, I compared feature points extracted from the 3D model with those measured by a total station. For a set of 20 feature points, the mean horizontal error was 1.9 cm and the mean vertical error was 2.5 cm. Additionally, I compared road cross-section widths: two sections measured by total station gave widths of 13.89 m and 14.05 m, while the 3D model gave 13.90 m and 14.04 m, with differences around 1 cm, all within acceptable tolerances.
Table 4 presents a comparison of work efficiency between China UAV oblique photogrammetry and traditional manual total station survey for the same project area.
| Comparison Item | China UAV Oblique Photogrammetry | Traditional Manual Survey |
|---|---|---|
| Field operation time (h) | 2 | 15 |
| Field personnel (persons) | 2 | 3 |
| Office time (h) | ||
| 3D modeling | 4.5 | – |
| 3D drawing | 1.5 | – |
| Topographic map drawing | – | 2 |
| Deliverables | DLG, 3D model, point cloud, orthophoto, etc. | DLG only |
Note: Field operation time includes GCP layout, topographic measurement, and flight operations. The China UAV method significantly reduced field time and personnel, while producing richer data products with comparable accuracy.
4. Conclusion
Municipal road engineering is a critical infrastructure for urban development, and accurate surveying directly impacts construction quality and efficiency. My experience with this case project demonstrates that China UAV oblique photogrammetry offers a highly efficient, accurate, and safe alternative to traditional ground-based methods. The key application points—thorough preparation, optimal flight route planning using a grid pattern, proper GCP layout with uniform distribution, two-pass aerial triangulation with control points, and rigorous accuracy verification—are essential to achieving reliable results. The technology not only speeds up field data acquisition but also generates comprehensive 3D models, point clouds, and orthophotos that facilitate design, construction, and maintenance phases. As China UAV technology continues to evolve, its integration with artificial intelligence and real-time kinematic positioning will further enhance the capabilities of oblique photogrammetry in smart city and digital twin applications. I recommend that surveyors and engineers in municipal road projects adopt China UAV oblique photogrammetry as a standard practice, while paying close attention to the specific requirements of each site. Future research may focus on automatic flight planning algorithms for dense urban areas and fusion of China UAV data with ground laser scanning for even higher precision. The results of this study confirm that China UAV oblique photogrammetry is a powerful tool that meets the rigorous demands of modern municipal engineering surveying.