PE&RS June 2002

Direct Georeferencing

Direct Georeferencing at The Ohio State
University: A Historical Perspective 

This article is the first in a series focused on The Ohio State University's (OSU) role in the development of direct georeferencing - the backbone of Mobile Mapping technology, over the past ten years. Special emphasis is placed on early developments and technological advancements in relation to land-based and airborne applications of this technique at OSU only and the research and development done in other establishments will be presented separately in the direct georeferencing column in the next few months. The details of the mobile mapping technology implementation and operational aspects can be found in the references. Another historical account is given in the PE&RS November 2001 column.

The Early Days of Direct Georeferencing: The Concept and its Evolution

The origin of direct georeferencing, also referred to as direct sensor orientation or direct orientation, dates back to the early 1990s, when the Center for Mapping (CFM) at OSU initiated the development of the GPSVan™, the first mobile mapping system (Bossler et al., 1991; Bossler, 1992; He et al., 1994b and c; Novak and Bossler, 1995). The Center for Mapping was established in June 1986 and launched with a NASA 5-year, $5 million grant, heavily leveraged with government, industrial, and university funds. The Center was established to serve as a focal point for interdisciplinary research, graduate studies and service within OSU in the field of geosciences, and to facilitate the government-industry cooperation for commercial benefits.

From the start, the Center focused its attention on the development of new technologies, starting with real-time environmental monitoring, spatial digital data processing, real-time mapping, and feature extraction from remotely sensed imagery, all in collaboration with a number of OSU faculty. The synergistic cooperation between the CFM and other research entities across the OSU campus, especially the Department of Geodetic Science and Surveying (DGSS) as well as strong ties to government and industry, created a rich environment for any interdisciplinary research involving high-precision navigation and spatial data acquisition and processing. In particular, the CFM and DGSS focused on promoting collaborative research in digital photogrammetry, multi-sensor system integration, and intelligent database design. In 1989 the Center prepared a proposal "The Application of Global Positioning System for Transportation Planning," which was sponsored jointly by the Federal Highway Administration, 38 State Departments of Transportation and Alberta, Canada, and the Defense Mapping Agency (currently NIMA). "The goal of this project," as stated in the CFM 1990 Annual Report, was "to build a mobile workstation that can automatically map and record crucial features of the nation's highways." Moreover, the road centerline information collected by the system could be processed into base maps and provide a cost effective foundation for GIS.

From the inception of this project, it was clear that the primary objective of providing improved quality spatial data in a short turnaround time could only be achieved through digital imaging sensor fusion, supported by robust positioning via GPS/inertial navigation integration for direct image georeferencing. Smart sensor/data fusion, along with the creation of a common frame to co-register multiple image data in a consistent manner, could support various mapping, assessment, inventory, change detection, targeting, object location and area classification, and other decision making tasks.

So, the concept of Mobile Mapping was born, and in 1990 the GPSVan™ design and successful implementation began. The primary driving force behind the idea of the automated and fully digital highway infrastructure data acquisition was the economy: "time and money are dissipated for mapping highway features with our present approaches," John Bossler, the CFM director, wrote in the 1991 CFM Annual Report, and he pointed out the increasing demand at the state and local levels for digital road data, which obviously were not available in the early 1990s. In the same document, a successful implementation of the first Mobile Mapping System (MMS) was reported. A Mobile Mapping System can be defined as a moving platform, upon which multiple sensor/measurement systems have been integrated to provide three-dimensional near-continuous positioning of both the platform and simultaneously collected geo-spatial data, with no or limited ground control using single or multiple GPS base stations.

A GPSVan™ prototype was designed for economic and rapid acquisition, analysis and processing of transportation data, based on integrated GPS, direction and distance measurement sensors, and video recording equipment. The system was intended to operate at highway speeds, and make metric measurements using a stereovision system in post-processing mode. In its initial implementation, completed in July 1991, the GPSVan™ was based on kinematic differential carrier phase and range data, and a Dead Reckoning System (DRS), providing accuracy of about one meter. DRS included a Three Axis Reference System (TARS) and wheel sensors (odometers). The TARS was based on a vertical gyro and a directional gyro in a self-contained gyro package, providing heading and attitude data. Odometers and gyroscopes were used to determine the distance traveled and the stereovision system's orientation, as well as to bridge the losses of GPS signal lock. The stereovision system consisted of two CCD cameras, supplemented by two analog Super-VHS video cameras, used for attribute identification (Novak, 1992; He and Novak, 1992). In addition, for the road attribute capture, the system used icons, programmed into a color touch screen. During the first two years (1990-1991) over 100 surveys were completed in several states, primarily to demonstrate the system's performance and promote the new technology. Several modifications to the hardware and the software resulted from these tests, namely, mounting an additional wheel counter at the left front wheel, eliminating the geometric bias caused by one wheel, integration of additional sensors to collect attributes, refinement of the data collection procedures, improved integration of GPS and inertial data, and software optimization rendering a 50 percent reduction in processing times.

April 1992 saw a new step in the modernization and further spreading of mobile mapping technology into commercial use: corporate partners were identified and affiliated with The Ohio State University Research Foundation, and the manufacturing of two second-generation systems began. The second generation of the GPSVan™ was based on dual frequency differential GPS, which enhanced the positioning quality and reliability. At the same time, the Center initiated its cooperation with NAVSYS, who was awarded a NASA SBIR award for the development of an inertial navigation system that could be integrated with a GPS-based navigation system, and thus directly advance the GPSVan™ development. Concurrently, the development of a comprehensive and automated image analysis and management system, fully integrated into the state-of-the-art GIS was initiated, based on the initial implementation in the GPSVan™ prototype. The new stereovision system was based on two 1280´1024 Kodak DCS cameras, replaced in 1994 by the 768x484 Pulnix TM-9700 model, and subsequently with 768´484 Pulnix TMC-9700 (color), allowing for collection of stereo image pairs in 15-meter intervals while traveling at 55 mph. (Details on the GPSVan™ imaging system can be found in He et al., 1994a and b; Toth, 1995a; Blaho and Toth, 1995; Bossler and Toth, 1996).

In addition, significant software enhancements were implemented in the GPSVan™ navigation post-processing module: a rigorous combination of the GPS-derived coordinates and the inertial data was implemented through the extended Kalman filter and a nonlinear smoother, allowing for calibration of the orientation errors and drifts in the inertial system, and semi-continuous positioning at 20 Hz rate. This facilitated the final implementation of the GPSVanTM positioning module (Da and Dedes, 1995). With these enhancements, the positioning accuracy reached the level of 5-10 cm (10-15 cm in vertical direction), when five or more satellites were available. GPS data outages of 30, 60, and 120 s caused the positioning degradation to the level of 0.2, 0.4, and 1.0 m, respectively (Bossler and Toth, 1995; Mallet and Dedes, 1995; Da and Dedes, 1995; Grejner-Brzezinska, 1996). Digital stereo pairs were processed in a post-mission mode, using the StereoMAP software (Novak, 1992; He and Novak, 1992), with the relative object coordinate accuracy reaching ~10 cm within 10-40 m from the vehicle (naturally, the absolute positioning was dictated by GPS, as explained above). StereoMAP went through several improvements over the next two years; see for example, He et al. (1994a and b), Novak et al. (1994), Bossler and Toth (1995), Toth (1995a and b). With these new initiatives and software/hardware enhancements, the GPSVan™ project was ready to market its second generation of mobile mapping technology.

In 1993 the new technology was used in a variety of projects. The most prominent examples are rectification of tax assessment maps for county auditors, development of base maps for a county engineer's office, and mapping a new subdivision for an electric utility. The GPSVan™ was also introduced to railroad inventory surveys, and provided "high quality" data to update Conrail maps of the railroad right-of-way. The GPSVan™ was mounted on high rails and operated along railroad tracks, which provided an extraordinarily cost effective way to map the track. At the same time, the GPSVan™ was licensed to Global Vision, Herndon, Virginia, and subsequently in 1994 to General Railway Signal Corporation (GRS) of Rochester, New York (exclusive rights to use GPSVan™ technology for rail and mass transit applications are under the trade name of TransVision™), and LAMBDA Tech International of Waukesha, Wisconsin (exclusive rights to use GPSVan™ technology in Thailand and seven other Southeast Asian countries is under the trade name of GPSVision™). Finally, in July 1994, TransMap, Inc., a CFM spin-off company was formed in Columbus, focusing its efforts on further commercialization of the GPSVan™ technology.

The most important years for GPSVan™ technology were 1994-1995. Now licensed for commercial use, it had to prove its applicability in production environments. And it passed with flying colors! Over 6,000 miles of Burlington Santa Fe Corporation (BNSF) railroad were surveyed, and the comparison with GPSVan™ versus ground truth revealed that over 95% of the quality checkpoints, the coordinates computed by the GPSVan™ were within 50 cm of the ground truth - an unprecedented accomplishment in both the accuracy and the number of miles surveyed. The GPSVan™ proved to be faster, better, and less costly compared to other mapping techniques. The third generation of GPSVan™, under the trade name of ON-SIGHT™ (Figure 1), is a turnkey system, providing high positioning accuracy of the features (with multiple attributes) extracted semi-automatically from color stereo imagery, using the STEPS™ (STEreo Positioning System) software (Habib, 1998; Habib et al., 1999). The imaging system consists of up to four color Pulnix TMC-9700 cameras, and the georeferencing is accomplished by the integration of dual frequency GPS with a strapdown inertial navigation system, LN-200, providing the feature positioning accuracy of 20-30 cm horizontally with good GPS geometry. However, since the accuracy strongly depends on the quality of GPS, which often times is compromised in the highway environments, the sub-meter accuracy is normally achieved (Novak and Nimz, 1997; http://www.transmap.com/). The primary applications are currently infrastructure management, law enforcement and accident reporting, engineering and planning, all at low cost and with a short turnaround time. Mobile mapping technology continues its quest throughout the world, and new-generation systems are built every year.

In summary, OSU has pioneered the concept of mobile mapping and continues its involvement in the development of new generations of georeferencing systems for direct orientation for both land-based and airborne applications, supporting a number of modern imaging sensors. In the mid-1990s, OSU experimented with airborne systems, and has developed a high-accuracy integrated GPS/INS navigation system for direct orientation of various airborne sensors. The initial implementation was based on "loose integration," and was developed for the US Navy, to support ground-penetrating radar, designed to locate and identify unexploded ordnance at weapons testing sites. The more advanced system, based on "tight" GPS/INS integration was developed later, for direct support of modern imaging sensors, such as high-resolution frame and line CCD, LIDAR, or multi/hyperspectral cameras. The system, called AIMS™ (Airborne Integrated Mapping System), was sponsored by a NASA grant, and industrial partners, Litton Guidance and Control Systems, Lockheed Martin Fairchild Semiconductors, and Trimble Navigation Ltd., and will be the subject of the next article in the series on "Direct Georeferencing at The Ohio State University: A Historical Perspective."

Acknowledgments
The author wants to thank Dr. John Bossler, the first director of the Center for Mapping, for reviewing this article.

References
Blaho, G., and Toth, C.: Field Experiences with a Fully Digital Mobile Stereo Image Acquisition System, Proc. ASPRS-ACSM Annual Convention, Vol. 2, pp. 97-104, 1995.

Bossler, J., Goad, C., Johnson, P. and Novak, K.: GPS and GIS Map the Nation's Highways, Geo Info Systems, Vol. 1, No. 3, pp. 27-37, 1991.

Bossler, J.: GPS Van: Input to GIS, Proceedings, ION GPS, pp. 427-437, 1992.

Bossler, J. D., and Toth, C.: Accuracies Obtained by the GPSVan™, Proceedings, IS/LIS'95, Vol. 1., pp. 70-77, 1995.

Bossler, J., and Toth, C.: Feature Collection by the GPSVan™, Analog Video or Stereo Digital Video?, Proceedings, ASPRS-ACSM Annual Convention, Vol. 2, pp. 108-115, 1996.

Da, R., and Dedes, G.: Nonlinear Smoothing of Dead Reckoning Data, Proceedings, Mobile Mapping Symposium, OSU, pp. 173-182, 1995.

Grejner-Brzezinska, D. A.: Positioning Accuracy of the GPSVanTM, Proceedings, 52nd Annual ION Meeting, Boston, June 19-21, pp. 657-665, 1996.

Habib, A., Uebbing, R., Novak, K.: Automatic Extraction of Road Signs from Color Terrestrial Imagery, Journal of Photogrammetric Engineering and Remote Sensing, Vol. 65, pp. 597-601, 1999.

Habib, A.: Motion Parameters Estimation by Tracking Stationary Three Dimensional Straight Lines in Image Sequences, International Journal of Photogrammetry and Remote Sensing, 53, pp. 174-182, 1998.

He, G.P. and Novak, K.: Automated Analysis of Highway Features from Digital Stereo-Images, International Archives of Photogrammetry and remote Sensing, Vol. XXIX, Part B3, 1992.

He, G.P., Novak, K., and Tang, W.: The Accuracy of Features Positioned with the GPSVan, ISPRS Commission II Symposium, Vol. 30, Part 2, pp. 480-486, 1994a.

He, G.P., Dedes, G., Orvets, G., and Bossler, J.D.: Generation of transportation GIS by Integrating GPS, INS and Computer Vision Technology, Proceedings, 3rd International Colloquium of LIESMARS, WTUSM, Wuhan, PR China, pp. 91-99, 1994b.

He, G.P., Cunningham, D. and Bossler, J.: Spatial Data Collection with the GPSVan Mobile Mapping System, Proceedings, ISPRS Commission IV Symposium, Vol. 30, Part 4, pp. 107-113, 1994c.

Mallett, A., and Dedes, G.: Real-Time On-The-Fly Ambiguity Resolution for cm-Level GPS Positioning, Proceedings, Mobile Mapping Symposium, OSU, pp. 115-122, 1995.

Novak, K.: Global Positioning with a Stereo-Vision System, Proceedings, 6th International Geodetic Symposium on Satellite Positioning, Columbus, OH, Vol. 2, pp. 702-711, 1992.

Novak, K., DaSilva, J., Toth, C.: Electronic Imaging Systems Developed at the Ohio State University, Proceedings, ISPRS Commission I Symposium, Vol. XXX, Part 1, pp. 56-61, 1994.

Novak, K., and Bossler, J.D.: Development and Application of the Highway Mapping System of Ohio State University, Photogrammetric Record, 15(85), pp. 123-134, 1995.

Novak, K. and Nimz, J.: GIS/GPS: Transportation Infrastructure Management, EOM Magazine, Vol. 6, No. 9, pp. 24-27, September 1997.

Schwarz, K.P., 1995. Integrated Airborne Navigation Systems for Photogrammetry, Photogrammetric Week'95 (in D. Fritsch and D. Hobbie, editors) Wichmann Verlag, Heidelberg, Germany, pp. 139 - 153.

Schwarz, K.P., M.A. Chapman, M.E. Cannon and P. Gong, 1993. An Integrated INS/GPS Approach to The Georeferencing of Remotely Sensed Data, PE&RS, 59(11): 1167-1674.

Toth, C.: Experiences with a Fully Digital Image Acquisition System, Proceedings, ASPRS-ACSM Annual Convention, Vol. 2, pp. 18-24, 1995a.

Toth, C.: A Conceptual Approach to Imaging for Mobile Mapping, Proceedings, Mobile Mapping Symposium, Columbus, OH, pp. 19-27, 1995b.

Edited by Dr. Mohamed M R Mostafa, Applanix Corporation

Top

Home