State-of-the-art large-format airborne imaging – evaluation of cameras and software

Duration: 6 months
Completion: July 2024
Supervisor & Examiner: Dr.-Ing. Michael Cramer

Changes in technical design of large format digital cameras, after 20 years of operational use, may affect the established process chains, which demands fur­­­t­her investigations. Unlike the current in-use camera system for national mapping (UltraCam Eagle Mark 3 by Vexcel Imaging), the state-of-the-art large-format aerial imaging sensor systems, including UltraCam Eagle 4.1 by Vexcel Imaging, DMC-4 by Leica Geosystems and UrbanMapper-2 by IGI, are designed with C­MOS sensors and different motion compensation methods to address image blur. These camera systems also employ different technologies- pan-sharpening and Bayer pattern techniques to capture color images. The features of the four camera systems are listed as the following table.

Additionally, the thesis evaluates the performance of aerial triangulation and digital surface model reconstruction by the latest photogrammetric software, ArcGIS Reality Studio. ArcGIS Reality Studio offers an efficient and easy-to-use protocol for aerial imaging evaluation employing GNSS/IMU-supported bundle adjustment. For analysis on object point accuracy of aerial triangulation, Match-AT modules in Trimble Inpho suite are utilized. Trimble Inpho suite is a widely implemented photogrammetric processing software tool suite developed by Trimble Geospatial and Match-AT offers bundle block adjustment with customized parameters.

Camera systems

The geometric and radiometric performance of the aforementioned camera systems is evaluated using the latest dataset acquired by AdV in the test site Neumünster. The empirical aerial datasets are collected between March and June in 2023. The test flights are taken with GSD 10cm as 3 strips covering the test field and designed as an 80% forward overlap and a 50% cross overlap. For all the four camera systems, the side overlap of all the four flights are approximately 50% , but the coverage in the side-track direction varies due to the differences in sensor size.

The aerial triangulation is carried out in Trimble inpho Match-AT modules and DSM is generated by nFrames SURE. With the experiments on the various GCP selection and a-priori standard deviation, 15 GCP that evenly distributed in the test field are selected and higher weights are assigned to X/Y coordinates of the object points.

The optical resolution and radiometry test field is designed to be located close to the center of the test field. The image sharpness is assessed by MTF10 measurement of the Siemens stars- that arranged in the radiometry test field- in the acquired images. The difference in the image sharpness by the four camera systems can be linked to various reasons such as the motion compensation method, color filtering or the image pre-processing as well as the environment components.

Software Evaluation

The photogrammetric software evaluation employs the same AdV dataset that was obtained by the UltraCam Eagle Mark 3. Aerial imaging geometric processing is conducted in Trimble Inpho suite version 12 in conjunction with nFrames SURE and ArcGIS Reality Studio (RS) version 23.1.0 for a comparison. These procedures are described as Project-A and Project-B, respectively, in the evaluation statement.

Regarding the DSM products, the height difference between given object points and sampled points on the surface model is calculated for both products. Also, the comparison of generated DSM and reference LiDAR data in different classes are performed.

The evaluation of the object point accuracy for all four products demonstrated that the precision of aerial triangulation was deemed satisfactory. Moreover, the radiometry performance of the 3 state-of-the-art camera systems generally represents considerable sharpness in the images. However, further investigation into the camera's radiometric resolution, specifically the relationship between radiometric performance observed at different partitions of the image, may be warranted to fully exploit the potential capabilities of large-format aerial imaging.

For aerial triangulation, ArcGIS Reality Studio presents a straightforward workflow and offers clear and intuitive results through its GUI-based visualization. ArcGIS RS extracts more than twice as many feature-based automatic tie points in its image alignment compared to Match-AT and relatively of a higher time consumption. By comparison, ArcGIS Reality Studio delivers comparable precise bundle adjustment results. Compared to Match-AT toolkit in Trimble Inpho suite, the mature aerial triangulation package, Arc-GIS Reality Studio delivers comparable precise bundle adjustment results. Still, to accommodate the diverse accuracy requirements of aerial imaging users, it could be suggested to implement a greater flexibility to customize the parameter configuration in the aerial triangulation process.