Camillo Ressl

Digital Photogrammetric Camera Evaluation - Generation of Di- gital Elevation Models

Summary: During the implementation of the DGPF-project on Digital Photogrammetric Camera Evaluation a team “Digital Elevation Models” was established. The main goal was to use the test’s framework for documentation and evaluation of the current state-of-the-art on photogrammetric 3D data capture from automatic image matching. During these investigations the accuracy and reliability of DSM rasters and 3D point clouds as derived from imagery of digital photogrammetric camera systems were evaluated. For this purpose they were compared to reference measurements from ground truth and airborne LiDAR. In addition to the evaluation of standard products, the usability of elevation data from image matching was investigated while aiming at specific applications in the context of urban modeling and forestry. Zusammenfassung: Wahrend des DGPF-Projektes zur Evaluierung digitaler photogrammetrischer Luftbildkamerasysteme wurde auch eine Auswertegruppe fur die Bewertung der Genauigkeit der Hohenmodellgenerierung etabliert. Dabei sollte der DGPF-Test genutzt werden, um den derzeitigen Stand der Technik der photogrammetrischen 3D Erfassug mittels automatischer Bildzuordnung zu dokumentieren. Hierfur wurden DSM Raster und 3D Punktwolken aus Bildern der photgrammetrischen Kamerasysteme abgeleitetet und die Qualitat dieser Ergebnisse in Bezug auf Genauigkeit und Zuverlassigkeit bewertet. Dabei wurde ein Vergleich zu terrestrischen Referenzmessungen und flugzeuggestutzen LiDAR Daten durchgefuhrt. Neben der qualitativen Bewertung von Standardprodukten wurde auch die Nutzbarkeit der Hohendaten fur spezielle Anwendungen beispielsweise im Kontext der 3D Stadmodellierung und Forstwirtschaft untersucht.

DGPF-Project: Evaluation of Digital Photogrammetric Camera Systems - Geometric Performance

The geometric performance of digital airborne cameras also including the impact of direct sensor orientation has been evaluated by a test of the German Society of Photogrammetry, Remote Sensing and Geoinformation (DGPF). This test includes following airborne photogrammetric cameras: the large format frame cameras Z/I Imaging DMC, Vexcel Imaging UltraCamX and the line scanning camera system Leica Geosystems ADS40 (2 generation) and Jena Optronik JAS-150 as well as the mid-format camera Rolleimetric AIC-x1 and the combination of four mid-format cameras Quattro-DigiCAM. The results presented in this paper were achieved by a group of researchers from different institutions, working independently from each other and with different programs for data acquisition and bundle block adjustment. Moreover, different adjustment configurations (i.e. with/without use of perspective centre coordinates and/or attitude information from GPS/inertial systems), and also different control point configurations have been used in the test; this results in a wide range of solutions and accuracy results which are not easy to compare, on the other hand this just shows the spectrum of possible solutions in operational applications.

Undistorting the Past: New Techniques for Orthorectification of Archaeological Aerial Frame Imagery

Archaeologists using airborne data can encounter a large variety of frame images in the course of their work. These range from vertical aerial photographs acquired with very expensive calibrated optics to oblique images from hand-held, uncalibrated cameras and even photographs shot with compact cameras from an array of unmanned airborne solutions. Additionally, imagery can be recorded in one or more spectral bands of the complete optical electromagnetic spectrum. However, these aerial images are rather useless from an archaeological standpoint as long as they are not interpreted in detail. Furthermore, the relevant archaeological information interpreted from these images has to be mapped and compared with information from other sources. To this end, the imagery must be accurately georeferenced, and the many geometrical distortions induced by the optics, the terrain and the camera tilt should be corrected. This chapter focuses on several types of archaeological airborne frame imagery, the distortion factors that are influencing these two-dimensional still images and the necessary steps to compute orthophotographs from them. Rather than detailing the conventional photogrammetric orthorectification workflows, this chapter mainly centres on the use of computer vision-based solutions such as structure from motion (SfM) and dense multi-view stereo (MVS). In addition to a theoretical underpinning of the working principles and algorithmic steps included in both SfM and MVS, real-world imagery originating from traditional and more advanced airborne imaging platforms will be used to illustrate the possibilities of such a computer vision-based approach: the variety of imagery that can be dealt with, how (accurately) these images can be transformed into map-like orthophotographs and how these results can aid in the documentation of archaeological resources at a variety of spatial scales. Moreover, the case studies detailed in this chapter will also prove that this approach might move beyond current restrictions of conventional photogrammetry due to its applicability to datasets that were previously thought to be unsuitable for convenient georeferencing.

Influence of surface reflectivity on reflectorless electronic distance measurement and terrestrial laser scanning

Abstract The uncertainty of electronic distance measurement to surfaces rather than to dedicated precisionre flectors (reflectorless EDM) is afected by the entire system comprising instrument, atmosphere and surface. The impact of the latter is significant for applications like geodetic monitoring, high-precision surface modelling or laser scanner self-calibration. Nevertheless, it has not yet received sufficient attention and is not well understood. We have carried out an experimental investigation of the impact of surface reflectivity on the distance measurements of a terrestrial laser scanner. The investigation helps to clarify (i)whether variations of reflectivity cause systematic deviations of reflectorless EDM, and (ii) if so, whether it is possible and worth modelling these deviations. The results show that differences in reflectivity may actually cause systematic deviations of a few mm with diffusely re- flecting surfaces and even more with directionally reflecting ones. Using abivariate quadratic polynomial we were able to approximate these deviations as a function of measured distance and measured signal strength alone. Using this approximation to predict corrections, the deviations of the measurements could be reduced by about 70% in our experiment.We conclude that there is a systematic effect of surface reflectivity (or equivalently received signal strength) on the distance measurement and that it is possible to model and predict this effect. Integration into laser scanner calibration models may be beneficial for high precision applications. The results may apply to a broad range of instruments, not only to the specific laser scanner used herein.

Integrated Change Detection and Classification in Urban Areas Based on Airborne Laser Scanning Point Clouds

This paper suggests a new approach for change detection (CD) in 3D point clouds. It combines classification and CD in one step using machine learning. The point cloud data of both epochs are merged for computing features of four types: features describing the point distribution, a feature relating to relative terrain elevation, features specific for the multi-target capability of laser scanning, and features combining the point clouds of both epochs to identify the change. All these features are merged in the points and then training samples are acquired to create the model for supervised classification, which is then applied to the whole study area. The final results reach an overall accuracy of over 90% for both epochs of eight classes: lost tree, new tree, lost building, new building, changed ground, unchanged building, unchanged tree, and unchanged ground.

Range calibration for terrestrial laser scanners and range cameras

Range cameras and terrestrial laser scanners provide 3D geometric information by directly measuring the range from the sensor to the object. Calibration of the ranging component has not been studied systematically yet, and this paper provides a first overview. The proposed approaches differ in the object space features used for calibration, the calibration models themselves, and possibly required environmental conditions. A number of approaches are reviewed within this framework and discussed. For terrestrial laser scanners, improvement in accuracy by a factor up to two is typical, whereas range camera calibration still lacks a proper model, and large systematic errors typically remain.

Impact of the Acquisition Geometry of Very High-Resolution Pléiades Imagery on the Accuracy of Canopy Height Models over Forested Alpine Regions

This work focuses on the accuracy estimation of canopy height models (CHMs) derived from image matching of Pléiades stereo imagery over forested mountain areas. To determine the height above ground and hence canopy height in forest areas, we use normalised digital surface models (nDSMs), computed as the differences between external high-resolution digital terrain models (DTMs) and digital surface models (DSMs) from Pléiades image matching. With the overall goal of testing the operational feasibility of Pléiades images for forest monitoring over mountain areas, two questions guide this work whose answers can help in identifying the optimal acquisition planning to derive CHMs. Specifically, we want to assess (1) the benefit of using tri-stereo images instead of stereo pairs, and (2) the impact of different viewing angles and topography. To answer the first question, we acquired new Pléiades data over a study site in Canton Ticino (Switzerland), and we compare the accuracies of CHMs from Pléiades tri-stereo and from each stereo pair combination. We perform the investigation on different viewing angles over a study area near Ljubljana (Slovenia), where three stereo pairs were acquired at one-day offsets. We focus the analyses on open stable and on tree covered areas. To evaluate the accuracy of Pléiades CHMs, we use CHMs from aerial image matching and airborne laser scanning as reference for the Ticino and Ljubljana study areas, respectively. For the two study areas, the statistics of the nDSMs in stable areas show median values close to the expected value of zero. The smallest standard deviation based on the median of absolute differences (σMAD) was 0.80 m for the forward-backward image pair in Ticino and 0.29 m in Ljubljana for the stereo images with the smallest absolute across-track angle (−5.3◦). The differences between the highest accuracy Pléiades CHMs and their reference CHMs show a median of 0.02 m in Ticino with a σMAD of 1.90 m and in Ljubljana a median of 0.32 m with a σMAD of 3.79 m. The discrepancies between these results are most likely attributed to differences in forest structure, particularly tree height, density, and forest gaps. Furthermore, it should be taken into account that temporal vegetational changes between the Pléiades and reference data acquisitions introduce additional, spurious CHM differences. Overall, for narrow forward–backward angle of convergence (12◦) and based on the used software and workflow to generate the nDSMs from Pléiades images, the results show that the differences between tri-stereo and stereo matching are rather small in terms of accuracy and completeness of the CHM/nDSMs. Therefore, a small angle of convergence does not constitute a major limiting factor. More relevant is the impact of a large across-track angle (19◦), which considerably reduces the quality of Pléiades CHMs/nDSMs. Remote Sens. 2018, 10, 1542; doi:10.3390/rs10101542 www.mdpi.com/journal/remotesensing Remote Sens. 2018, 10, 1542 2 of 22

Roughness Spectra Derived from Multi-Scale LiDAR Point Clouds of a Gravel Surface: A Comparison and Sensitivity Analysis

The roughness spectrum (i.e., the power spectral density) is a derivative of digital terrain models (DTMs) that is used as a surface roughness descriptor in many geomorphological and physical models. Although light detection and ranging (LiDAR) has become one of the main data sources for DTM calculation, it is still unknown how roughness spectra are affected when calculated from different LiDAR point clouds, or when they are processed differently. In this paper, we used three different LiDAR point clouds of a 1 m × 10 m gravel plot to derive and analyze the roughness spectra from the interpolated DTMs. The LiDAR point clouds were acquired using terrestrial laser scanning (TLS), and laser scanning from both an unmanned aerial vehicle (ULS) and an airplane (ALS). The corresponding roughness spectra are derived first as ensemble averaged periodograms and then the spectral differences are analyzed with a dB threshold that is based on the 95% confidence intervals of the periodograms. The aim is to determine scales (spatial wavelengths) over which the analyzed spectra can be used interchangeably. The results show that one TLS scan can measure the roughness spectra for wavelengths larger than 1 cm (i.e., two times its footprint size) and up to 10 m, with spectral differences less than 0.65 dB. For the same dB threshold, the ULS and TLS spectra can be used interchangeably for wavelengths larger than about 1.2 dm (i.e., five times the ULS footprint size). However, the interpolation parameters should be optimized to make the ULS spectrum more accurate at wavelengths smaller than 1 m. The plot size was, however, too small to draw particular conclusions about ALS spectra. These results show that novel ULS data has a high potential to replace TLS for roughness spectrum calculation in many applications.

Evaluation of the elevation model influence on the orthorectification of Sentinel-2 satellite images over Austria

ABSTRACT The European Space Agency (ESA) is using PlanetDEM for the generation of their Sentinel-2 Level-1C orthophotos. There are concerns that this DEM is not accurate enough for this purpose, especially over mountainous regions like in Austria. In this study, we investigate the height accuracy of PlanetDEM in Austria by comparing it with a reference DEM that is derived from airborne laser scanning. Afterwards, we predict the errors in the Sentinel orthophotos that are caused by these DEM errors: 99.8% of all pixels have a PlanetDEM-induced displacement between ±10 m (i.e. one Sentinel-2 pixel) for all investigated four satellite tracks over Austria. Still, at a few spots, the location errors can reach 60 m. ESA’s goal is to achieve a multi-temporal geo-registration quality (for Level-1C products) of 3 m (95.5% confidence) once the Global Reference Image is established. If we assume that this error budget consists partly of a georeferencing and partly of a DEM-induced error and both are equal, then the DEM-induced error may only amount to 2 m. In this case, PlanetDEM would not allow for this accuracy in the mountainous regions, because there, according to the predictions of this study, 95.5% of the errors are between ±3.8 m.

Relative orientation of videos from range imaging cameras

In this paper we investigate the determination of camera relative orientation in videos from time of flight (ToF) range imaging camera. The task of estimating the relative orientation is realized by fusion of range flow and optical flow constraints, which integrates the range and the intensity channels in a single framework. We demonstrate our approach on videos from a ToF camera involving camera translation and rotational motion and compare it with the ground truth data. Furthermore we distinguish camera motion from an independently moving object using a robust adjustment.