Using mobile laser scanning system in geodetical routines

Introduction

In the last 20 years instead of the methods of the traditional geodesy, the current solutions are the methods of the geoinformatics. The computer engineering has taken the lead. Nowadays the maps are made by computer. The information technology determines the directives and also the possibilities. In our company (Geodézia Kft.) we used to apply 3D laser scanners for years both the static and the mobile devices. These tools are capable of the accurate and detailed surveying of the environment and also capable of the geometrical and the visual data collection. Therefore we use these devices not only by the traditional geodetical challenges, but we are searching for other potential applications. The mobile laser scanning system of the Geodézia Kft. is a complex solution that combines the digital photogrammetry, the 3D laser scanner and the mobility. This system has different type of sensors which are capable continuously determination the path of the vehicle and simultaneously capable of acquisition spatial data. These complementary spatial data is based on the different type of the sensors. We have stabilized positioning devices (GNSS, IMU, DMI), digital cameras and 3D laser scanner on a car. While driving, the vehicle records the route and the street view in geometrically evaluable method. (1. figure) This system can be stabilized on different type of vehicles (ship, railway vehicles, etc.).

1. figure: System components

System components

Positioning system (GNSS/IMU)

The positioning system is the priority part of the mobile mapping system. After the positioning, every measured attribute gets into the database (to the geoinformatics system). The main function of the positioning is to precisely determine the position of the mobile vehicle and the system. This positioning is mainly based on GNSS (GPS, GLONASS). The relevant limitation of the GNSS routines is the measuring accuracy and the measuring speed. The accuracy can be increased by using permanent stations and virtual basis. The speed can be increased by using complementary systems. The INS (Inertial Navigation System) is a system what contains the DMI (Distance Measuring Instrument), the IMU (Inertial Measurement Unit) and supported by software. It is based on the GPS/INS NOVATEL SPAN platform. The total accuracy is actually the accuracy of the GNSS/IMU system. 

Based on our experiences the available accuracy with post-processing is 3-5 cm. In case of the error value is bigger than 5 cm after the post-processing, it is necessary to determine pass points. By choosing the pass points the most important is that the chosen points have to be well identifiable in the point-cloud and also in the surface. Such a point can be for example a corner of a road marking, or a corner of a manhole cover, etc.

 

2. figure: Camera system with pointcloud

Camera system

Our camera system consists of six cameras. Each of them guarantees high resolution colorful images. Their resolutions are 1600 x 1200 (1,92 megapixel). The cameras locate round on the top of the car. Their pictures are partly overlapping each other so during the post-processing these images can be interconnected and then they can be used to spatial processing. The profit of the stereo cameras is that spatial coordinates can be calculated by the formulas defined in photogrammetry. That means that beside the point-cloud we get a photogrammetric product, which provides significant assistance during the processing. (Image 2) Making images is based on a previously defined distance, but this can be modified while driving. Considering the image resolution and the distance of the image making, the system generates 1 gigabyte data within one kilometer. The system has a 500 Gb storage capacity so more daily data collection can be recorded. The automatic processing system gives the opportunity to display the images in 360 degrees, so the user can visualize panorama pictures which provide better orientation and processing.

Laser component

The laser scanner installed our system is an ultra-high speed phase based Leica scanner. This scanner was bought earlier for other purposes, mainly for static tasks (scanning constructions in/outside). This scanner can be separated from the system, therefore we can use it as a static instrument. The scanner produces a new point-row in every movement during mobile application. The distance of these point-rows (they are actually plans) depends on the speed of the car and the efficiency of the scanner. The speed of our scanner is 1 million point/sec – beside 50 Hz rotation speed. Range: 187 m. Relative distance measuring accuracy is less than 1 mm (till 25 m). After touring the relevant path there and back, with these parameters the scanner can make an adequate accuracy point-cloud for the processing. The scanner was fixed 40 degree tilted on the car. This guarantees that the scanner sees more side of the building. It collects data not from only two, but three sides.

Surveying and utilization

The safety and economy surveying based on the premise of the route-planning. By the office phase should be informed about the circumstances of the surveying area. It is useful to select a place where the instrument can be fixed on/off and installed. The system activation is consists of a static and a dynamic initialization. It takes 15 minutes and needs to make it under the open sky, then the system components can work in synch and the system can acquire the appropriate GNSS position. During the system are working, a managing and a processing workstation helps the operator. The user interface runs under the Windows operation systems. It informs the user uninterrupted from each component in different windows. Then we can get information about the number of the satellites and the GPS/GNSS parameters. Also here can be done the manual recording and sign the things with coordinates. The program is able to handle the uploaded maps to relieve the navigation. The program allows the camera operation depending on the lighting conditions and also the image recording different from the pre-defined operation. The data collection ensures the multi-functional utilization thanks to the technology. The mobile mapping system is attached with high accuracy the whole 3D image of the cities and towns. From this point-cloud can be determined the situation of the objects (for example public utilities) with geodetical accuracy. Because every object was mapped, then the editor will be able to process other objects from the point-cloud. With this technology the user can check efficiently and fast the status of the roads and the public utilities belonging them. The status of the railway infrastructure can be also checked fast and simple with 3D scanning. It helps us to determine where precisely the objects are which are necessary to change or service.

3. figure Railway spatial image from pointcloud

By high-resolution scanning the natural and synthetical (road, loess wall, dam, etc.) objects, the user can get a lot of information from the point-cloud. This information helps us to precisely evaluate our environment, and to forecast the natural disasters. Up to the present our works were mainly connected to the public utility surveying and to the propose mapping. But we were made so detailed surveying that later it can be used to other purposes (for example to make a high accuracy 3D model). 

Point-cloud processing

During the laser scanning survey, the collected data are not sorted like in the traditional in the field surveying. The operator can determine in the field that which objects, ground point, other things he/she wants to survey. In this method, the effective data collection happens in the office (like in the photogrammetric processing). The point-clouds consists of millions of points, therefore the traditional editor software (MicroStation, ITR, AutoCAD) can not handle them, or just with huge compromises. We have tested some software created to point-cloud processing. A lot of them are such applications which were written for a well-known CAD (or another geoinformatics) software. These programs have some advantages: the software they based on use quite popular graphic file formats, their editor functions are advanced. But they have some disadvantages: the user need to buy more software together. For every single application, it is necessary to buy quite expensive base software. Besides that, the user also has to buy the point-cloud processing software, which price is almost as expensive as the basic software. After we tried some software, we have chosen the PointCloudScene (PCS) software from the DIGIKART Kft. This is an independent software, so the user does not need an other software to run the PCS. The PCS handles the well-known point-cloud file formats: LAS, bin (TerraScan), fls (FARO), PTS. The user can process in DXF, ESRI shape, or Personal Geodatabase (mdb) format. The ESRI formats have the advantage that any number of attributes can connect to the geometric data. Therefore the PCS is able to upload a geoinformatics database. This software is able to visualize several point-clouds together. First the loaded point-clouds can be visualized in the 2D view and the editor can determine which part of the point-cloud will be visualized in 3D. (4. figure)

4. figure: Selecting a poincloud section for display

Usually the editor uses two different type of processing. In the first case, in the 2D view (top view) the editor can draw the line geometry objects (building, fence, etc.) and the well-identified point geometry objects (pillar, tree, etc.). Then in the 3D view the editor completes and specifies these objects elevation. In the other case, the whole editing happens in the 3D view. If the hardware tools (3D graphic card, 3D monitor, 3D glasses) are convenient, the user can process in 3D. Based on our experiences when the end product will be in 2D, then the processing is faster in 2D. But if the objects elevation is relevant, then the second method is more effective. The points can be visualized according to the elevation or intensity in the 2D/3D view. In case of colorized point-cloud, of course the visualization also can be happened due to the points own color. The different type of the visualization helps the processing. By using intensity colorization, the processing of the manhole cover, sewer, etc. will be easier, because the laser beam rebounds with different intensity from the different material surface. The PCS supports the point classification due to different properties (intensity, height. etc.). For example it is able to separate the points with 0-0,5 meter above and under the ground surface, therefore the user can easier process the objects on ground level, because at any time he/she can switch on/off the visibility of the interfering points (outside the ground level). For classification, a classification toolbox is available, which contain some automatic and semi-automatic tools. In case of classified point-cloud there is an opportunity for the coloration and visualization due to these classes. During process a classified point-cloud, the points can be filtered quickly with a few click. For example thanks to the height filtering, all through a ground level processing (ditch, road surface, etc.) the higher objects (pillar, tree) do not disturb the operator. It is very useful by stereo-processing in the 3D view. The classification is time-consuming, therefore before each new project the user has to decide that the classification is profitable or not (all through the editing). Then the editor can execute the classification. There is an opportunity to edit point item, line item and polygon item. Text or block can be applied to the point items. The toggle plane can be visualized in any position. This helps the processing. This plane can be used for filtering/shrouding points in the relevant view. The biggest problem by the processing that the user can choose the point he/she wanted to edit from the millions of points. The different types of snap methods help to choose the relevant point. The spatial cross-hair is a useful tool in complex cases, which can be moved by mouse or it can be placed to specified coordinates. By using this cross-hair, the editor can measure at that place, where there was no point in the point-cloud, but for example an edge of a buttress has to be there (based on the environment). By clicking a mouse, a specific point can be selected, but the user has to check this point from other views (rotate, move) to make sure that this point is the point he/she wanted to select. By using the stereo visualization in 3D, these checking can be skipped. Based on our experiences not everybody can work continuous with this technology. The PCS is able to visualize the point-clouds and simultaneously the georeferenced images in case of mobile mapping system. The details which were not clearly identifiable (in the point-cloud) become clear thanks for visualizing the point-cloud and the image together. (5. figure)

5. figure:  Georeferenced images

The coordinates of the evaluable points can be calculated if it is necessary, by using the overlapping image-pair. After the processing, the PCS can export the shape files to DXF, or with using an adequate template file, the end product can be produced. Summarizing, the processing of the point-clouds produced by laser scanner is not a simple task. The point-cloud contains confusingly large amount of information. Out of the ordinary, the editor becomes to the surveyor, so he/she has to learn the “in the field-thinking”. The editor has to decide, which objects and how detailed will be represent in the vector file. Consequently the time spent with surveying decreases significant, but the processing in the office takes much more time. The latter can be planned and managed much better, and work in the office is not depend on the weather. If some objects had not evaluated for any reason, then it would be sufficient to open the point-cloud again and process the missing objects. Therefore it would not be necessary to go out and make additional measurements. Based on our experiences the additional surveying is not necessary, except if there is a shrouded area in the point-cloud. Based on our experiences the time required for the point-cloud processing is continuously decreasing during the same type of works and the editors become more practiced. Considering the advance of the processing software, the day when the point-cloud processing will take no more as the traditional geodetic processing isn’t very far. 

The video presented the mobile mapping system can be seen in YouTube. The title of the video is: “Mobile Mapping System - 3D mobil laserscanning - Geodezia Ltd.”.

 

 
References
  1. Barsi Árpád: Integrált és mobil térképezés.. Elektronikus segédlet MSc képzéshez, Budapest, 2011.
  2. Gombás László: Leica-Geosoft mobil térképező rendszer – új trendek a digitális téradatgyűjtésben. Geodézia és Kartográfia LXIV. évfolyam, 2012/1–2
  3. Use of Mobile Mapping System in Traditional Surveying Tasks
 
Authors note
Péter KUNFALVI
project leader
peter.kunfalvi@geodeziakft.hu
László KULCSÁR
project leader
kulcsi@geodeziakft.hu