Direct Georeferencing
LAND-BASED MOBILE MAPPING SYSTEMS By Cameron Ellum and Naser El-Sheimy — Canada This article, part of a series on direct georeferencing, focuses on land-based mobile mapping systems. In particular, a brief history of their development is reviewed, implementations of such systems are discussed, and an extensive list of references on implementations is given. The theory behind georeferencing, presented in earlier columns, is revisited, including a summary of error analysis.
History of Land-Based Mobile Mapping System The first operational land-based MMS was developed by the Center for Mapping at the Ohio State University. Their system called GPSVan integrated a code-only GPS receiver, two digital CCD cameras, two color video cameras and several deadreckoning sensors. All components were mounted on a van the GPS provided the position of the van and the images from the CCD cameras were used to determine the positions of points relative to the van (Goad, 1991; Novak, 1991). The deadreckoning sensors, which consisted of two gyroscopes and an odometer (wheel counter) on each of the front wheels, were primarily used to bridge GPS signal outages. These sensors were also used to provide orientation information for the exposure stations. The two video cameras were used solely for archival purposes and to aid attribute identification no relative positioning was performed using the video imagery. Using bundle-adjustment techniques with relative-orientation constraints, the GPSVan was able to achieve relative object space accuracies of approximately 10 cm. Unfortunately, because only carrier-smoothed code-differential GPS was used, absolute object-space accuracies were limited to 1-3 m. It is worth noting that GPS was the primary motivator for the development of both land-based and airborne MMS, and the absolute accuracies are still largely dependant upon the performance of the GPS. Indeed, except for some rare indoor MMS, such as that described in El-Hakim et al. (1997), all MMS currently use GPS for positioning. GPSVan successfully illustrated how land-based multi-sensor systems could improve the efficiency of GIS and mapping data collection. However, the absolute accuracy of the object space points was too poor for many applications. Also, the dead reckoning sensors in the GPSVan were not very suitable for bridging GPS outages. Therefore, further developments of landbased mobile mapping systems focused on improving system reliability and increasing absolute object space accuracies. One obvious technique for improving absolute accuracy was the use of carrier phase differential GPS. Similarly, the obvious choice for a more accurate dead-reckoning sensor was high precision Inertial Measurement Units (IMUs). In both regards, the development of land-based MMS again followed that of their airborne counterparts, as the potential of integrating IMUs with carrier-phase differential GPS for aerial photography had been identified as early as the mid 1980s (Goldfarb, 1985). The use of an IMU has an additional advantage over other types of dead-reckoning sensors since it also provides high-accuracy orientation information for the exposure stations. Further dePHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
velopments of land-based MMS based on GPSVan or involving its researchers including NAVSYS GPS/Inertial Mapping system (GIM) and LambdaTechs GPSVision all used IMUs as their dead-reckoning sensors (Coetsee and Brown, 1994; He et al., 1996). Later independent implementations of landbased MMS added dual-frequency carrier-phase differential GPS, more accurate IMUs, and more sophisticated processing techniques examples of some later systems include the VISAT system (Schwarz et al., 1993), KiSS (Hock et al., 1995), and GI-EYE (Brown, 1998). The VISAT system in its final form was notable because of the large number of imaging sensors it employed. Where previous land-based MMS were simple stereovision systems employing only two forward facing cameras, VISAT had eight cameras permitting more flexible data collection and better imaging geometry. As a final note, the commercial viability of land-based MMS is evident in the number of successful spin-off companies created by the original researchers. For example, GPSVan and its research spawned two companies Transmap Corp. and Lambda Tech International (Transmap, 2001; Lambda Tech, 2001). Analytical Surveying, Inc. is also successfully operating the VISAT van. In one important aspect land-based MMS have largely led their airborne counterparts. Namely, land-based MMS have, from their inception, used digital cameras as their imaging sensors. This was possible because of the much smaller camerato-object distances in land-based MMS when compared to airborne systems. The poor resolution of CCD chips meant that they could not be used in aerial applications without a noticeable accuracy degradation. Indeed, the resolution of CCD chips has only recently improved to the level that they can be used in airborne mapping systems. The use of digital cameras is advantageous because they eliminate the requirement to scan photographs. Consequently they substantially reduce the period from raw data collection to extracted data dissemination. A list of some land-based MMS is presented in Table 1. Not considered in the table or, indeed, in this article are systems that merely mount a navigation sensor on a moving platform. Such systems face the same limitations as traditional landbased surveying systems namely the necessity to occupy each point of interest. Furthermore, such systems are also significantly less appropriate for GIS data collection because of the requirement to manually enter attribute information. Also not included again, in the table are systems that merely use GPS with a camera exclusively for archival purposes. Such systems do not use the imagery for positioning purposes, and are consequently not mobile mapping systems.
Coordinate Determination from Georeferenced Platforms MMS integrate navigation sensors and algorithms together with sensors that can be used to determine the positions of points remotely. All the sensors are rigidly mounted together on a platform; the former sensors determine the position and orientation of the platform, and the latter sensors determine continued on page 15
January 2002
13
Direct Georeferencing continued from page 13
the position of points external to the platform. The sensors that are used for the remote position determination are predominantly photographic sensors and thus they are typically referred to as imaging sensors (El-Sheimy, 1999). However, additional sensors such as laser rangefinders (Li et al., 1999) or laser scanners are also used in MMS and therefore the more general terms of mapping sensors (Li, 1997) or relative sensors (Novak, 1995) may also be used when referring to the remote sensors. In the following imaging, mapping, relative and remote sensors are used interchangeably. Evident in the discussion of the history of land-based MMS is that the platform that holds the sensors is typically a van. However, the use of other platforms, such as trains (Blaho and Toth, 1995; Sternberg et al., 2001) and even people (Barker-Benfield, 2000; Ellum and El-Sheimy, 2001), has also been investigated and implemented. The strength of Mobile Mapping Systems lays in their ability to directly georeference their mapping sensors. A mapping sensor is georeferenced when its position and orientation relative to a mapping coordinate frame is known. Once georeferenced, the mapping sensor can be used to determine the positions of points external to the platform in the same
mapping coordinate frame. In the direct georeferencing done by MMS the navigation sensors on the platform are used to determine its position and orientation. This is fundamentally different from traditional indirect georeferencing where the position and orientation of the platform are determined using measurements made to control points. These control points are established through a field survey prior to or after data acquisition, and their establishment is typically expensive and time-consuming. Therefore, eliminating this step results in obvious decreases in both the cost and time-requirements for data collection. The task of establishing ground control is additionally complicated since its cost and time requirements are frequently difficult to estimate. Also, for many terrestrial surveys the establishment of sufficient control is virtually impossible for example, consider the control requirements to map an entire city using close-range photogrammetry. Finally, for some mapping sensors such as laser-scanners or push-broom CCD arrays it is difficult or even impossible to establish control. The use of these sensors is not practical unless directgeoreferencing is performed. The basis for all direct georeferencing formulas is a sevencontinued on page 16
Table 1: Examples of Land-Based Multi-Sensor Systems
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
January 2002
15
Direct Georeferencing continued from page 15
parameter conformal transformation where the coordinates of a point in the MMS imaging sensors coordinate frame are related to their coordinates in a mapping coordinate frame . (1) In the above equation, is the position of the mapping and are sensor in the mapping coordinate frame, and respectively the scale factor and rotation matrix between the mapping sensor coordinate frame and the mapping coordinate frame. In an MMS, all three parameters are typically measured albeit indirectly. In actuality, the position of a GPS antenna on the platform, and the orientation of an IMU or other attitude-sensing device are measured. The scale factor can be determined directly for example, from a laser rangefinder - or indirectly for example, using stereo-techniques with two images. , and are typiBecause direct measurements of , cally not made, Equation 1 is normally extended to include terms that account for the indirect measurements. Also, MMS are moving during or between measurements. Consequently, the position and orientation of the system with respect to the mapping coordinate system are changing with time, and therefore Equation 1 must be modified to reflect this. As an example of both of these changes, the georeferencing formula for a system integrating a mapping sensor with incorporating GPS and an IMU is
Figure 1 shows the development of this equation, and Table 3 describes the meaning and determination of the various parameters. By examining this table, it can be seen that there are many quantities dependant upon either calibration or synchronization, and thus total system accuracy critically depends on both of these tasks being correctly performed. To examine the effects of calibration, synchronization, or measurement errors, a first order error analysis can be performed by linearizing the georeferencing equation and adding a term for the synchronization error. For example, the error analysis for Equation (2) is
Performing an error pre-analysis (i.e., prior to construction of an MMS) is an important step; such an analysis is required in order to choose the appropriate components that allow the system to meet the desired accuracies and satisfy the operational parameters. (Continued on Page 17)
(2)
16
January 2002
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
Direct Georeferencing This technique yields the most accurate object-space coordinates.
ò
Cameron Ellum and Naser El-Sheimy are with The University of Calgary, Department of Geomatics Engineering, Canada Edited by Mohamed Mostafa, Applanix Corporation References Alexander, J.F. 1996. Gator Communicator: Design of a Hand Held Digital Mapper. Proceedings of Third Congress on Computing in Civil Engineering. Anaheim, CA. June 17-19, 1996. pp. 1052-1057. Barker-Benfield, S. 2000. Extra dimension: Professor-Patented Mapping Device Combines Old, New. The Florida Times-Union. Website: http:// www.jacksonville.com/tu-online/stories/071200/ bus_3519070.html. Accessed 15 Jan 2001.
Figure 1: Development of Georeferencing formula
Benning, W. and T. Aussems. 1998. Mobile Mapping by a Car Driven Survey System (CDSS). Proceedings of the Symposium on Geodesy for Geotechnical and Structural Engineering. H. Kahmen, E. Brückl & T. Wunderlich, eds. International Association of Geodesy (IAG). April 2022, 1998. Eisenstadt, Austria.
Table 3: Terms in Expanded Georeferencing Formula
Blaho, G. and C. Toth. 1995. Field Experiences with a Fully Digital Mobile Stereo Image Acquisition System. Proceedings of 1995 Mobile Mapping Symposium. May 24-26, 1995. Columbus, OH. pp. 97-104. Brown, A. 1998. High Accuracy Targeting Using a GPSAided Inertial Measurement Unit. Proceedings of the 54th Annual Meeting. Institute of Navigation (ION). June, 1998. Denver, CO. Coetsee, J., A. Brown, and J. Bossler. 1994. GIS data Collection Using the GPSVan Supported by a GPS/Inertial mapping System. Proceedings of GPS-94. Institute of Navigation (ION). September 20-23, 1994. Salt Lake City, UT.
It should be noted that the position and orientation are typically determined using a previously integrated GPS and IMU. In this case, Equation 2 reduces to (4) Of course, the missing calibration terms from Equation 2 have not disappeared they are merely hidden in the term, which is the position of the IMU determined from the integrated GPS/IMU navigation system. The optimal integration of navigation sensors particularly GPS and IMU is currently a subject of much research interest both inside and outside of the mobile mapping community. It is typically done using a Kalman filter that estimates the errors in the position, velocity, orientation of the integrated system, and the errors in the accelerometer and gyroscope biases in the IMU. Finally, it is worth noting that Equation 1 is the basis of the collinearity equations used in a photogrammetric bundle adjustment. This implies that the estimates of exposure station position and orientation can be input as weighted parameter observations into an adjustment using unified least squares. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
El-Hakim, S.F., P. Boulanger, F. Blais, and J.A. Beraldin. 1997. A System for Indoor 3-D Mapping and Virtual Environments. Proceedings of SPIE Vol. 3174 - Videometrics V. The International Society for Optical Engineering (SPIE). San Diego, CA. July 1997. pp.21-35. Ellum, C.M. and N. El.Sheimy. 2001. A Mobile Mapping System for the Survey Community. Proceedings of The 3rd International Symposium on Mobile Mapping Technology (MMS 2001). Cario, Egypt. January 3-5, 2001. On CD-ROM. El-Sheimy, N. and K.-P. Schwarz. 1999. Navigating urban areas by VISAT - A Mobile Mapping System Integrating GPS/INS/Digital Cameras for GIS Applications. Navigation. Institute of Navigation (ION). Vol. 45. No. 4. pp. 275-285. El-Sheimy, N. 1999. Mobile Multi-sensor Systems: The New Trend in Mapping and GIS Applications. Geodesy Beyond 2000: The Challenges of the First Decade. International Association of Geodesy Symposia Volume 120. Springer-Verlag Berlin. pp. 319-324. Goad, C.C. 1991. The Ohio State University Mapping Syscontinued on page 28
January 2002
17
continued from page 27
ner specifies a geographic polygon, flight line bearing, and photo scale, the XMap PhotoFlight application looks at the digital elevation data to determine the altitude and placement of each flight line to ensure the proper side-to-side overlap and scale. After the flight lines are set, the trigger points for the photos are calculated. The elevation data is again used in the calculation to ensure that the proper overlap along the flight line is achieved. It includes features to help users easily generate block and corridor flight plans, mark project boundaries using advanced draw tools, or import ShapeFiles directly to the map. Each flight plan produces files containing DIRECT GEOREFERENCING continued from page 17
tem: The Positioning Component. Proceedings of the 47th Annual Meeting. The Institute of Navigation (ION). June 10-12. Williamsburg, VA. pp. 121-124. Goldfarb, J.M. 1985. Exposure Station Control for Aerotriangulation with an INS-Differential GPS. Proceedings of Inertial Technology for Surveying and Geodesy. Banff, Canada. September 16-20. pp. 777-789. Graefe, G., W. Caspary, H. Heister, J. Klemm and M. Sever. 2001. The Road Data Acquisition System MoSES - Determination and Accuracy of Trajectory Data Gained with the Applanix POS/LV. Proceedings of The 3rd International Symposium on Mobile Mapping Technology (MMS 2001). Cario, Egypt. January 3-5, 2001. On CD-ROM. He, G., G. Orvets, R. Hammersley. 1996. Capturing Urban Infrastructure Data using Mobile Mapping System. Proceeding of the 52nd Annual Meeting. The Institute of Navigation (ION). June 19-21, 1996. Cambridge, MA. pp. 667674. Hock, C., W. Caspary, H. Heister, J. Klemm, and H. Sternberg. 1995. Architecture and Design of the Kinematic Survey System KiSS. Proceedings of the 3rd International
28
January 2002
flight line details for CCNS and trigger points for the camera system. For information, contact 1-800-795-3951 or visit the DeLorme Web site at www.delorme.com/ photoflight. Eastman Kodak Company has released their new IDL 5.5 software. IDL (Interactive Data Language) software is Kodaks world-class data visualization and analysis software. IDL software was created by Research Systems Inc. (RSI), a wholly owned subsidiary of Eastman Kodak Company. Version 5.5 delivers new features that increase performance and flexibility, including built-in multi-threading capabilities, new gridding and interpolation methods and new context senWorkshop on High Precision Navigation. Stuttgart, Germany. April, 1995. pp. 569-576. Lambda Tech. 2001. Welcome to Lambda Tech International. Website: www.lambdatech.com. Accessed 10 Feb. 2001. Li, D. S.-D. Zhong, S.-X. He, and H. Zheng. 1999. A Mobile Mapping System Based on GPS, GIS and Multi-Sensor. Proceedings International Workshop on Mobile Mapping Technology. Bangkok, Thailand. April 21-23, 1999. pp. 13-1 - 1-3-5. Li, Q., B. Li, J. Chen, Q. Hu, and Y. Li. 2001. 3D Mobile Mapping System for Road Modeling. Proceedings of The 3rd International Symposium on Mobile Mapping Technology (MMS 2001). Cario, Egypt. January 3-5, 2001. On CDROM. Li, R. 1997. Mobile Mapping: An Emerging Technology for Spatial Data Acquisition. Photogrammetric Engineering and Remote Sensing (PE&RS). Vol. 63. No. 9. pp. 1085-1092. Novak, N. 1991. The Ohio State University Mapping System: The Stereo Vision System Component. Proceedings of the 47th Annual Meeting. The Institute of Navigation (ION). June 10-12. Williamsburg, VA. pp. 121-124. Novak, K. 1995. Mobile Mapping
sitive widgets (pop-up menus). Additionally, version 5.5 makes IDL software an ActiveX software container as well as an ActiveX software control, allowing the user to embed components such as Microsoft Excel spreadsheets in IDL software applications. One of the major new enhancements is the addition of multithreading capability, which has been added to IDL software for the Windows and Unix platforms. For information, contact www.kodak.com/go/globalimaging.
Websites The National Remote Sensing and Space Law Center has a new website. Please see www.spacelaw.olemiss.edu
Technology for GIS Data Collection. Photogrammetric Engineering and Remote Sensing (PE&RS). Vol. 61. No. 5. pp. 493-501. Reed, M.D., C.E. Landry, and K.C. Werther. 1996. The Application of Air and Ground Based Laser Mapping Systems to Transmission Line Corridor Surveys. Proceedings of Position, Location and Navigation Symposium (PLANS 1996). Institute of Electrical and Electronics Engineers (IEEE). Atlanta, GA. April 22-26. pp. 444-451. Schwarz, K.-P., H.E. Martell, N. ElSheimy, R. Li, M.A. Chapman, and D. Cosandier. 1993. VISAT - A Mobile Highway Survey System of High Accuracy. Proceeding of the Vehicle Navigation and Information Systems Conference. Institute of Electrical and Electronics Engineers (IEEE). Ottawa, Canada. October 12-15. pp. 467-481. Sternberg, H., W. Caspary, H. Heister, J. Klemm. 2001. Mobile Data Capturing on Roads And Railways Utilizing the Kinematic Survey System KiSS. Proceedings of The 3rd International Symposium on Mobile Mapping Technology (MMS 2001). Cario, Egypt. January 3-5, 2001. On CD-ROM. Transmap. 2001. Welcome to TRANSMAP Corp.. Website: www.transmap.com. Accessed 10 Feb. 2001. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING