Petri Rönnholm

Calibration of Laser-derived Tree Height Estimates by Means of Photogrammetric Techniques

Techniques based on laser point clouds and digital terrestrial images were demonstrated for the calibration of tree-height estimation. Individual tree heights can be roughly estimated from laser scanning data by using the approximated ground level and the highest hit of the treetop. However, laser-derived measurements often underestimate tree heights. This underestimation can arise from various error sources. Digital terrestrial images can be used to verify and understand the behaviour of laser point clouds. When laser data are backprojected in a close-range image, it is possible to show where each laser beam has reflected. This, however, requires a proper orientation of the images. In this study an interactive orientation method was used to derive image orientations, using one laser strip at a time as the reference data. Consequently, the backprojection of laser point clouds confirmed the height underestimations found by comparing the tacheometer reference measurements with the laser-derived tree heights. In addition, by using the described procedure the cause of underestimating tree heights could be explained.

Map updating and change detection using vehicle-based laser scanning

The vehicle-based laser scanning (VLS, also known as mobile mapping) is a new technology, which is currently under development for creating 3D models of the surrounding environment. VLS is based on the integration of GPS, IMU, laser scanner and preferably digital cameras mounted on top of a moving platform, i.e. a car in most applications. VLS is a logical development after the first operative Airborne Laser Scanner (ALS) in 1994 and Terrestrial Laser Scanners mounted on top of a tripod. The data/image processing of VLS are mainly based on modifications of the methods created for ALS and TLS taking into account the differences of VLS compared to ALS and TLS. Compared to ALS, the geometry of VLS scanning is different and the pulse density varies as function of range. Two main differences between stationary TLS and constantly moving VLS are the evenness of the data and the perspective. In VLS, the point cloud is evenly distributed along the driving direction, and the viewing direction to the target remains constant. In the stop-and-go mode, the data characteristics of the VLS and conventional TLS are similar. A reasonable amount of research has been done to develop methods for single-time VLS processing, but there have not been any attempts to our knowledge of multitemporal processing of VLS data. In this paper, the high potential of change detection based on multitemporal VLS point clouds was demonstrated. Example cases include the change detection of city models and defoliation of city trees. A method to map biomass and biomass change of (city) trees was developed.

Orientation of Airborne Laser Scanning Point Clouds with Multi-View, Multi-Scale Image Blocks

Comprehensive 3D modeling of our environment requires integration of terrestrial and airborne data, which is collected, preferably, using laser scanning and photogrammetric methods. However, integration of these multi-source data requires accurate relative orientations. In this article, two methods for solving relative orientation problems are presented. The first method includes registration by minimizing the distances between of an airborne laser point cloud and a 3D model. The 3D model was derived from photogrammetric measurements and terrestrial laser scanning points. The first method was used as a reference and for validation. Having completed registration in the object space, the relative orientation between images and laser point cloud is known. The second method utilizes an interactive orientation method between a multi-scale image block and a laser point cloud. The multi-scale image block includes both aerial and terrestrial images. Experiments with the multi-scale image block revealed that the accuracy of a relative orientation increased when more images were included in the block. The orientations of the first and second methods were compared. The comparison showed that correct rotations were the most difficult to detect accurately by using the interactive method. Because the interactive method forces laser scanning data to fit with the images, inaccurate rotations cause corresponding shifts to image positions. However, in a test case, in which the orientation differences included only shifts, the interactive method could solve the relative orientation of an aerial image and airborne laser scanning data repeatedly within a couple of centimeters.

Improving Automation in Map Updating Based on National Laser Scanning, Classification Trees, Object-Based Change Detection and 3D Object Reconstruction

Previously, several countries have performed countrywide laser scanning, but mainly for DTM purposes. This paper discusses the possibility to use countrywide collection of laser data, possibly multi-temporal laser data, for updating Topographic and forest databases, especially concerning the detection of the changed buildings or trees and reconstructing them from laser scanner data. Knowledge obtained in both the EuroSDR comparison of building extraction and EuroSDR/ISPRS test on tree extraction is applied in order to predict the obtainable quality. Some examples to use change detection are given. Rough concepts for implementation of the ALS surveys are depicted. The use of high-density cross-strips could be used for both strip adjustment and quality control. The intensity of the collected laser data is proposed to be calibrated. We also propose to use classification tree techniques together with existing methods in order to automatically classify laser point clouds.

3D city model for mobile phone using MMS data

Recently, research towards using 3D city models for personal navigation has been rapidly increasing. In this paper, an approach for 3D city model reconstruction for the application of mobile phone-based navigation is presented, which is based on data collected from vehicle-based mobile mapping system (MMS). Our method is performed based on three objectives: small model size, perfect accuracy control as well as good visual effect. Small model size is achieved by simplified object geometry and reduced texture resolution. Model accuracy is controlled by extracting building outlines from classified point cloud and overlapping with final 3D model. Model completeness is checked by comparing resulting model with original images. Good visual effect is realized by applying photo-realistic texture. Photorealistic texture provides rich information for the reconstructed 3D scene. By applying this approach, in test area, 3D city model is successfully reconstructed.

Evaluating the Correctness of Airborne Laser Scanning Data Heights Using Vehicle-Based RTK and VRS GPS Observations

In this study, we describe a system in which a GPS receiver mounted on the roof of a car is used to provide reference information to evaluate the elevation accuracy and georeferencing of airborne laser scanning (ALS) point clouds. The concept was evaluated in the Klaukkala test area where a number of roads were traversed to collect real-time kinematic data. Two test cases were evaluated, including one case using the real-time kinematic (RTK) method with a dedicated GPS base station at a known benchmark in the area and another case using the GNSSnet virtual reference station service (VRS). The utility of both GPS methods was confirmed. When all test data were included, the mean difference between ALS data and GPS-based observations was −2.4 cm for both RTK and VRS GPS cases. The corresponding dispersions were ±4.5 cm and ±5.9 cm, respectively. In addition, our examination did not reveal the presence of any significant rotation between ALS and GPS data.

RELATIVE ORIENTATION BETWEEN A SINGLE FRAME IMAGE AND LIDAR POINT CLOUD USING LINEAR FEATURES

Registration of multi-source remote sensing data is an essential task prior their efficient integrated use. It is known that accurate registration of different data sources, such as aerial frame images and lidar data, is a challenging process, where extraction and selection of robust tie features is the key issue. In the presented approach, we used linear features, namely roof ridges, as tie features. Roof ridges derived from lidar data are automatically located in the 2D image plane and the relative orientation is based on the well-known coplanarity condition. According to the results, the average registration (absolute) errors varied between 0.003 to 0.196 m in the X direction, between 0.018 to 0.282 m in the Y direction and between 0.010 to 0.967 m in the Z direction. Rotation (absolute) errors varied between 0.001 to 0.078 degrees, 0.006 to 0.466 degrees and 0.013 to 0.115 degrees for ω, ϕ and κ rotations, respectively. This study revealed that the method has potential in automatic relative orientation of a single frame image and lidar data. However, the distribution, orientation and the number of successfully located tie features have an essential role in succeeding in the task.

Radial Distortion from Epipolar Constraint for Rectilinear Cameras

Lens distortion causes difficulties for 3D reconstruction, when uncalibrated image sets with weak geometry are used. We show that the largest part of lens distortion, known as the radial distortion, can be estimated along with the center of distortion from the epipolar constraint separately and before bundle adjustment without any calibration rig. The estimate converges as more image pairs are added. Descriptor matched scale-invariant feature (SIFT) point pairs that contain false matches can readily be given to our algorithm, EPOS (EpiPOlar-based Solver), as input. The processing is automated to the point where EPOS solves the distortion whether its type is barrel or pincushion or reports if there is no need for correction.

FILTERING THE OUTLIERS FROM BACKPACK MOBILE LASER SCANNING DATA

Backpack mobile laser scanners provide rapid data acquisition and allow access to all walkable areas. We aim to remove outliers from the data of a prototype backpack mobile laser scanning system. After removing dark intensity points, our approach mainly uses voxel grids. Firstly, we remove points that only have a few points in the neighborhood. Secondly, we search for outlier clusters with a 3D convolution of voxel grids. Thirdly, outlier clusters located in the air or below the ground are removed with a single elimination bin. As a result, the majority of outliers are successfully removed from backpack mobile laser scanning data.

Development and implementation of a new Masters Programme in Geoinformatics at Aalto University, Finland.

A major renovation in teaching geoinformatics has been made at Aalto University, Finland. In this article, the development process and courses in the new Master in Geoinformatics are described from the perspective of photogrammetry and laser scanning. However, the new curriculum includes teaching in photogrammetry, laser scanning, remote sensing, geodesy, cartography, and geoinformation techniques, covering all aspects of the surveying sciences. The teaching language in all courses is English, allowing international students to participate. In addition, the new curriculum is part of the Nordic Master in Cold Climate Engineering. The students participating in this masters get a double degree. Aalto University grants the degree of Master of Science (Technology), and the Technical University of Denmark grants the degree of Master of Science in Earth and Space Physics and Engineering.